important note - coexist · 2019-10-31 · important note this coexist deliverable is part of a...

Important note

This CoEXist deliverable is part of a series of deliverables and reflects the state of development of the CoEXist project at the publication time (i.e. April 2018).

Therefore, some of the information presented in the present deliverable can be outdated. The final specification and designs will be presented in the deliverable

D4.3 which is expected to be published on early 2020.

For more information, you can visit our website (www.h2020-coexist.eu) or contact the CoEXist partners at: https://www.h2020-coexist.eu/contacts/

http://www.h2020-coexist.eu/

http://www.h2020-coexist.eu/

https://www.h2020-coexist.eu/contacts/

https://www.h2020-coexist.eu/contacts/

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 723201

Scenario

specifications for

eight use cases

Subtitle title goes here

Version: 1.0

Date: 24.04.19

Author: Johan Olstam and Fredrik Johansson

The sole responsibility for the content of this document lies with the authors. It does not

necessarily reflect the opinion of the European Union. Neither the EASME nor the European

Commission are responsible for any use that may be made of the information contained therein.


h2020-coexist.eu

Page 2 of 120 h2020-coexist.eu

Page 2 of 120

Document Control Page

Title Scenario specifications for eight use cases

Creator Johan Olstam

Editor Johan Olstam & Fredrik Johansson

Brief Description

Presents the scenarios under which the eight CoEXist use cases are to

be investigated. In CoEXist, the scenarios of a use case specify the

level of CAV introduction, travel demand, and potential traveller

adaptation, and to what extent and how these aspects should be

varied.

Reviewers Bernard Gyergyay and Syrus Gomari

Contributors

Johan Olstam, Fredrik Johansson, Mikael Ivari, Lina Svensson, Nina

Galligani Vardheim, Frank van den Bosch, Brian Matthews, Ammar

Anvar, John Miles, Gisa Gaietto, Markus Friedrich, Jörg Sonnleitner,

Adriano Alessandrini, Peter Sukennik

Type (Deliverable/Milestone) Deliverable

Format

Creation date 2018-02-07

Version number 1.0

Version date 2018-04-24

Last modified by Johan Olstam

Rights

Audience Internal

Public

Restricted, access granted to: EU Commission

Action requested To be revised by Partners involved in the preparation of the

Deliverable

For approval of the WP Manager

For approval of the Internal Reviewer (if required)

For approval of the Project Co-ordinator

Deadline for approval


h2020-coexist.eu


Page 3 of 120

Version Date Modified by Comments

0.1 2018-02-07 Johan Olstam First draft

0.2 2018-03-04 Fredrik Johansson,

Charlotte Fléchon & Johan

Olstam

First revision + first draft of

summary of Stuttgart use cases

0.3 2018-04-04 Fredrik Johansson &

Johan Olstam Final draft

0.3 2018-04-05 Syrus Gomari Added comments and edits

0.4 2018-04-12 Markus Friedrich CAV definitions

0.5 2018-04-16 Johan Olstam Revisions based on the review

comments

0.6 2018-04-19 Johan Olstam Finalisation

Table of contents

1 Introduction .................................................................................................. 6

1.1 Definition of a use case and scenario in the CoEXist context .............................................. 6

1.2 Aim ...................................................................................................................................... 7

1.3 Report structure ................................................................................................................... 7

2 Development process .................................................................................. 7

3 The scenario specification template ........................................................... 9

3.1 Level of automation and connectivity ................................................................................... 9

3.1.1 Connectivity functions ............................................................................................................. 10

3.1.2 AV classes.............................................................................................................................. 10

3.1.3 Driving logics .......................................................................................................................... 11

3.1.4 CAV-functions ........................................................................................................................ 11

3.2 Travel demand ................................................................................................................... 11

3.3 Traveller behaviour adaptation .......................................................................................... 12

3.4 Stages of coexistence ....................................................................................................... 12

3.5 Definition of road environments ......................................................................................... 12

3.6 Expectations on AV class driving logics for different road environments ........................... 13


h2020-coexist.eu


Page 4 of 120

3.7 Relevant combinations of CAV driving logics and functions .............................................. 15

4 Summary of scenario specifications ........................................................ 17

4.1 Gothenburg, Sweden ......................................................................................................... 17

4.1.1 Use case 1: Shared space (microscopic modelling) ................................................................ 17

4.1.2 Use case 2: Accessibility during long-term construction works (macroscopic modelling) ........ 18

4.2 Helmond, Netherlands ....................................................................................................... 18

4.2.1 Use case 3: Signalised intersection including pedestrians and cyclists (microscopic

modelling) ........................................................................................................................................... 18

4.2.2 Use case 4: Transition from interurban highway to arterial (microscopic modelling) ............... 19

4.3 Milton Keynes, UK ............................................................................................................. 19

4.3.1 Use case 5: Waiting and drop-off areas for passengers (microscopic modelling) ................... 19

4.3.2 Use case 6: Loading and unloading areas for freight (microscopic modelling) ........................ 20

4.4 Stuttgart, Germany ............................................................................................................ 21

4.4.1 Use case 7: Impacts of CAV on travel time and mode choice on a network level (macroscopic

modelling) ........................................................................................................................................... 21

4.4.2 Use case 8: Impact of driverless car- and ridesharing services (macroscopic modelling) ....... 22

5 Generic scenarios ...................................................................................... 22

5.1 Penetration rates ............................................................................................................... 22

5.2 Possible behavioural adaptation ........................................................................................ 23

6 Conclusions and lessons learnt ............................................................... 24

7 Partners ...................................................................................................... 26

Appendix A Driving Logics ............................................................................. 27

The “rail-safe” logic .................................................................................................................... 27

The “cautious driving” logic ........................................................................................................ 30

The “normal driving” logic ........................................................................................................... 31

The “all-knowing driving” logic .................................................................................................... 31

Appendix B Descriptions of AV-functions ..................................................... 33

Appendix C Use case 1: Shared space .......................................................... 35

Appendix D Use case 2: Accessibility during long-term construction

works……….. ..................................................................................................... 47

Appendix E Use case 3: Signalised intersection including pedestrians and

cyclists……… .................................................................................................... 57

Appendix F Use case 4: Transition from interurban highway to arterial..... 66


h2020-coexist.eu


Page 5 of 120

Appendix G Use case 5: Waiting and drop-off areas for passengers .......... 75

Appendix H Use case 6: Loading and unloading areas for freight .............. 88

Appendix I Use case 7: Impacts of CAV on travel time and mode choice on

a network level ................................................................................................ 101

Appendix J Use case 8: Impact of driverless car- and ridesharing

services…….. ................................................................................................... 110


h2020-coexist.eu


Page 6 of 120

1 Introduction

CoEXist address three key steps to achieve Automation-ready transport and road infrastructure

planning:

Automation-ready transport modelling: Develop a validated extension of existing microscopic

traffic flow simulation and macroscopic transport modelling tools to include various types of CAVs

(passenger cars/light-freight vehicles, automation levels).

Automation-ready road infrastructure: Create a tool to assess the impact of CAVs on traffic

efficiency, safety, and space demand and development of design recommendations for

Automation-ready infrastructure.

Automation-ready road authorities: Elaboration of eight use cases in four European local

authorities, to demonstrate the above tools and to develop concrete Automation-ready

infrastructure and policy action plans and recommendations for local authorities.

The extended traffic and transport models and the assessment tool will be demonstrated by applying

them to the eight use cases. There are several uncertainties regarding the introduction of CAVs. One

way of taking these uncertainties into account is to conduct scenario-based analysis. This report

describes the scenarios under which each of the eight CoEXist use cases are to be investigated.

1.1 Definition of a use case and scenario in the CoEXist context

In CoEXist, a use case is a traffic environment that should be investigated with respect to the

introduction of connected and automated vehicles (CAVs). The specification of each use case includes

descriptions of the present infrastructure, travel demand, travel patterns, traffic control, and traffic

conditions. The use case specification also includes questions and hypotheses connected to the

introduction of CAVs in the use case specific traffic environment. Furthermore, the specification includes

a draft list of potential measures (changes in infrastructure, traffic control, public transport, policy

measures, etc.) that might be interesting to investigate. Finally, the specification describes the available

models and data. Detailed specifications of the eight use cases in CoEXist is given in a separate

deliverable (D1.3).

The scenarios are meant to describe how each use case will treat uncertain factors related to

technological development and the reactions of the travellers to the new technology in terms of level of

automation, travel demand, and traveller behaviour adaptation. In CoEXist, the scenarios of a use case

specify the level of CAV introduction, travel demand, and potential traveller adaptation, and to what

extent and how these aspects should be varied in the simulations. There are several uncertainties

regarding the introduction of CAVs and not all can be taken into account in the scenarios. The choice of

these three types of uncertainty factors was made based on the discussions at the CoEXist scenario

workshop that took place 12th October 2017 in Brussels, Belgium. The focus is on uncertain factors

related to the modelling of traffic during the transition period. Uncertain factors related to for example

development and acceptance of shared services, car ownership, integrity, trust in authorities, attitudes

and acceptance of automated vehicles, etc. are not explicit included, but to some extent implicit included

via penetration rates and traveller behaviour adaptation.


h2020-coexist.eu


Page 7 of 120

1.2 Aim

The aim of this report is to present the scenarios under which the use cases are to be investigated. The

purposes of the completed template are:

To inform the reader of the assumptions made in the scenarios under which the use-case will be

investigated.

To provide input to the experimental design work to be performed in WP3.

To facilitate information exchange and cooperation between the cities and technical support

partners.

To facilitate dissemination of the CoEXist project.

1.3 Report structure

This report consists of two parts. The first part describes the process used to develop the detailed

scenario specifications (chapter 2), specifications of the different types of uncertain factors considered

(chapter 3), summaries of the scenarios for each use case (chapter 4), generalisations of the scenarios

for the eight use cases (chapter 5), and presents conclusions and lessons learnt (chapter 6). The first

part is accompanied by two appendices including definitions of the different driving logics (Appendix A)

and the different CAV-functions (Appendix B) authored by Adriano Alessandrini, Markus Friedrich and

Peter Sukennik. The second part consists of detailed specifications of scenarios for the eight use cases

(Appendix C - Appendix J). The first part is authored by Johan Olstam and Fredrik Johansson and the

scenario specification appendices in the second part are authored by the responsible city and support

partner and edited by Johan Olstam and Fredrik Johansson. The authorship for each use case is stated

in the beginning of each use case description in Appendix C - Appendix J.

2 Development process

The selection of use cases and specification of the scenarios are based on several discussion rounds,

among the CoEXist consortium partners and cities, about the practicality and fit with regards to the

specific context of each use case. The first drafts of the use cases were presented in the proposal. The

process for further specification of the use cases and the scenarios is described in Figure 1. To allow for

more detailed specification of the use cases that fulfil the aims and ensure consistent description, use

case and scenario specification templates were developed and circulated among the cities and their

support partners. First drafts of the use case specifications for the eight use cases were prepared and

used as input to a use case and scenario specification workshop in which the cities and the consortium

partners thoroughly discussed the aim, scope, relevance, feasibility in terms of modelling, potential

measures, uncertain factors, etc. The outcome of the workshop was then used by the cities and the

support partners to finalise the use case and scenario specifications.

In the next stage of CoEXist, the use case and scenario specifications will be used as a basis for

developing experimental designs for each use case, i.e. detailed specifications of baselines, measures

(e.g. changes to road design), travel demand, level of automation, road user adaptation, etc.. The

‘experimental design workshop’ will take place 15-16th May 2018 in Gothenburg, Sweden.


h2020-coexist.eu


Page 8 of 120

Figure 1 Flowchart describing the different activities in the development of the use cases and scenarios

Use case and scenario specification workshop

Development of use case template

Development of scenario template

Use case template

Scenario template

Use case specification – use

case 1


case 2


case 3


case 4


case 5


case 6


case 7


case 8

Scenario specification – use

case 1


case 2


case 3


case 4


case 5


case 6


case 7


case 8

Feedback from consortium (city and

support partners)

Feedback from consortium (city and

support partners)

Experimental design workshop


h2020-coexist.eu


Page 9 of 120

3 The scenario specification template

The completed scenario specification template is meant to describe how the use case will treat three

types of uncertain factors related to technological development and the reactions of the travellers to the

new technology in terms of:

Level of automation

Travel demand

Traveller behaviour adaptation

These three types of uncertain factors depend on the time horizon in terms of development, and

deployment of CAVs. In the scenario specification the time dimension is defined in terms of three

different stages of coexistence:

Introductory

Established

Prevalent

The uncertain factors can either:

Be held constant within a stage of coexistence (introductory, established, and prevalent)

(But might vary between different stages)

Varied in a range

(An approximate description of in which range the factors might vary)

It is important to remember that the number of scenarios increases dramatically for each uncertain factor

that is varied, both within and between stages. The scenario specification template is aimed to be a

support to make a relevant and balanced selection of which uncertain factors to vary and to what extent.

At this stage, only the approximate levels of the three types of uncertain factors need to be defined; the

exact values will be specified in Task 3.1 and the experimental design template and workshop.

The different types of uncertain factors are discussed in more detail in section 3.1 - 3.3 while the

description of different stages of coexistence is given in section 3.4.

3.1 Level of automation and connectivity

The level of automation is specified in three steps:

AV class (Basic AV, Intermediate AV and Advanced AV)

Driving logic (Rail-safe, Cautious, Normal, All-knowing) for different road environments

Functions (ACC, LKA, Jam assistance, etc.)

Each step is a further specification of the step above.


h2020-coexist.eu


Page 10 of 120

3.1.1 Connectivity functions

Connected vehicles are vehicles that exchange data with applications via the internet, with other vehicles

(V2V) or with the road infrastructure (V2I). Data exchange is not limited to AV as many conventional

vehicles also exchange data for information and control purposes. And AV may still be able to perform

certain driving functions, even in situations with failing data exchange.

To clarify the scope of connectivity within CoEXist three categories of connectivity are distinguished:

Exchange of data for vehicle information functions: This category of data exchange provides the

vehicle or the vehicle user with data on the network (road graph, static signs), the state of the

network (e.g. speeds, disturbances, construction sites) and the environment of the network (e.g.

POIs, weather). It also covers data the vehicle gives to service providers (e.g. vehicle position, state

of the vehicle). The data can be used for off-board navigation, fleet management, emergency

services (eCall) and the collection of traffic data. This category of data exchange can be found in

conventional and in automated vehicles.

Exchange of data for vehicle communication functions: This category of data exchange permits

communication between specific vehicles or specific roadside devices. Communication can be both

direct (V2V and V2I) or via the Internet. This function enables AV to better estimate the behaviour of

neighbouring vehicles (e.g. indicating a lane change) or to record the current or future state of a

traffic sign (traffic light, variable speed limits).

Exchange of data for vehicle cooperating functions: This category of data exchange permits an

AV to cooperate with other AV or traffic control devices. Cooperation includes merging, lane

changing and the forming of platoons.

CoEXist does not address the impact of data for vehicle information functions as this is a general topic of

ITS equally important for conventional and automated vehicles. Vehicle communication and cooperation

functions are considered in CoEXist as long as they concern the movement of individual AV. CoEXist

focuses on the vehicle and does not examine possible ways to improve the throughput of road facilities

resulting from an adaptive traffic control based on communication or cooperation between vehicles and

control devices.

3.1.2 AV classes

An AV class is an aggregate description of the behaviour and capabilities of the vehicles:

Basic AV: First generation of AVs with SAE1 level 4 capabilities only for one directional traffic

environments with physical separation with active modes. The behaviour is in general quite cautious

and risk minimizing. Basic AVs will not have dedicated devices for vehicle communication and

cooperating functions.

Intermediate AV: Second generation of AVs with level 4 capabilities in some road environments and

driving context. The behaviour at more complicated road environments and driving context is still

cautious and risk minimizing while the behaviour at less complicated road environments and driving

1 SAE International, 2016. SAE Standard J3016, Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road

Motor Vehicles.


h2020-coexist.eu


Page 11 of 120

context can be less cautious and still be safe. Intermediate AVs may have dedicated devices for

vehicle communication and cooperating functions, but are not depended on them.

Advanced AV: The third generation of AVs with level 4 capabilities in most road environments and

driving context. The behaviour and how cautious the behaviour is vary depending on road

environment and driving context. Advanced AVs will have dedicated devices for vehicle

communication and cooperating functions, but are not depended on them.

3.1.3 Driving logics

The AV class specifies which driving logic is used for each road type. The driving logics (see Appendix A

for a detailed description of the different driving logics) considered in CoEXist are:

Rail-safe: This is a simple logic based on the switch principle: if anything is on the collision course

OFF if not ON. The vehicle follows a pre-defined path for the whole trajectory. This logic does not

require specific devices for vehicle communication and cooperation.

Cautious: The cautious driving logic calculates gaps accurately and only merges when gaps are

acceptable, and it slows down every time its sensors can have blind angles to have no surprises.

This logic does not require specific devices for vehicle communication and cooperation.

Normal: The “normal” driving logic uses the logic of an average driver but with the augmented (or

diminished) capacities of the sensors for the perception of the surroundings. This type of driving logic

may require devices for vehicle communication and cooperation.

All-knowing: Perfect perception and prediction of the surroundings and the behaviour of the other

road users. This automated-driver is capable of forcing his way on other drivers whenever is needed

without however ever causing accidents. This type of driving logic requires devices for vehicle

communication. If the devices fail, the logic may fall back to a cautious driving logic. With additional

communication devices and control logics the driving logic enables cooperation with other CAVs.

3.1.4 CAV-functions

The last step is the specification of which functions and capabilities are expected to be relevant for the

use case, for each driving logic and road environment. This step could be specified for each vehicle but

here it is assumed that the driving logics of all vehicle types have identical functions (at least the ones

relevant for the scenarios at hand). Note, however, that different vehicle types still can have different

behaviours and capabilities on a specific road type, since the specification of AV class through

specification of driving logic on each road type is vehicle type specific. The specifications of the different

CAV-functions are given in Appendix B.

3.2 Travel demand

How the introduction of CAVs influence traffic performance, safety and space efficiency might depend on

the travel demand and the congestion level. Thus, it might be necessary to conduct traffic model

experiments for several demand configurations representing different time periods of the day, current

travel demand or prognosis demand levels.


h2020-coexist.eu


Page 12 of 120

3.3 Traveller behaviour adaptation

The basic assumption is that the behaviours of non-connected and non-automated road users are the

same as today. However, the introduction of CAVs could imply changes in the behaviour and interaction

of other road users (conventional vehicles, pedestrians, bicyclists, etc.) with CAVs and/or vehicles in

general. The basic assumption is that other road users interact with CAVs in the same way as they

interact with manually driven vehicles today. If these assumptions are valid or not are highly uncertain.

Thus, the behaviour of these other road users might need to be varied for some of the use cases.

3.4 Stages of coexistence

To investigate the range of conditions that are likely to occur during the gradual introduction of CAVs, a

use case may define three stages of coexistence: the introductory, established, and prevalent stages.

These should not be defined in terms of specific number of years in the future, but rather by the level of

automation in the use case. The use of three stages enables limitation of the number of scenarios, while

still providing insight into the whole range of the CoEXist period of introduction of CAVs and estimating

penetration levels at which there will be noticeable benefits from the introduction of CAVs into the urban

traffic mix. The exact nature of the stages may vary significantly between use cases, and all three stages

may not be relevant for all use cases, but the general characteristics of the stages are:

Introductory: Automated driving has been introduced, but the majority of vehicles are conventional

cars. Automated driving is in general significantly constrained by limitations (real or perceived) in the

technology.

Established: Automated driving has been established as an important mode in some areas.

Conventional driving still dominates some areas due to limitations (real or perceived) in the

technology.

Prevalent: Automated driving is the norm, but conventional driving is still present.

3.5 Definition of road environments

In the scenario specification template four types of road environments are used:

Motorway: Multi lane roads with physical barriers between directions and grade separated

intersections.

Arterial: Single or multilane roads with at grade intersections (mainly larger type of intersections as

signalized intersections or roundabouts). Bicycle and pedestrian traffic are clearly separated from the

vehicle traffic either by physical barriers or medians. Vehicles, bicycles and pedestrians interact at

intersections.

Urban street: Single or multi lane roads with at grade intersections (also stop or yield regulated

intersection). No clear separation between vehicle traffic and pedestrian and bicycle traffic.

Walkways and bikeways directly at the side of the vehicle lanes.

Shared space: Vehicles, bicycles and pedestrians share the same space, which can be

unstructured or semi-structured.


h2020-coexist.eu


Page 13 of 120

3.6 Expectations on AV class driving logics for different road environments

This section summarises the initial assumptions with respect to which type of driving logic that is used by

an AV-class, for the different types of vehicles (cars, trucks, buses, and minibus2) and four different types

of road environments (motorway, arterial, urban street, and shared space). Specified below are general

expectations that were used as a starting point in the scenario specification for each use case. These

expectations may be adjusted in the scenario specification for each use case, see Appendix C –

Appendix J.

Table 1 presents the expectations for cars. The first generation of AVs (Basic AV) are assumed to be

able to drive in automated mode only on motorways and arterials. Furthermore, they are assumed to

drive according to the cautious driving logic on these road types. The Intermediate AVs are assumed to

be able to use the normal driving logic on motorways due to development of sensor technology and

anticipation capabilities. The more complex arterials with at-grade intersections with interactions with

active modes still constrain the AVs capabilities and the behaviour is still according to the cautious

driving logic. Exceptions might be highly separated arterials with total conflict separation between

vehicles and active modes, in which Intermediate AVs can be assumed to be able to drive according to

the normal driving logic. Furthermore, the Intermediate AV is assumed to be able to drive according to a

cautious driving logic on urban streets and according to the rail-safe logic in shared spaces. However, it

is questionable to what extent drivers would accept the rail-safe logic driving in shared space due to the

high “politeness” of the driving logic and the resulting low speed. Thus, it might from a traffic simulation

point of view be more reasonable to assume manual driving. The Advanced AV class cars are assumed

to be able to drive in all the road environments but with more cautious driving logics for the more

complex road environments, ranging from the all-knowing driving logic on motorways and arterials to the

normal driving logic on arterials and cautious logic in shared spaces.

2 Minibuses (sometimes also called shuttles) with not necessarily seat-belted passengers for first and last mile services, e.g. NAVYA, 2getthere, and Ollie.


h2020-coexist.eu


Page 14 of 120

Table 1 Specification of the driving logics to be simulated for the three AV classes in the different road environments for the cars & trucks.

Vehicle type: Car & Truck

Road type Driving logic: (Rail-Safe (RS), Cautious (C), Normal (N), All-Knowing (AK), Manual (M))

Basic AV Intermediate AV Advanced AV

Motorway C N AK

Arterial C C / N3 AK

Urban street M C N

Shared space M RS4 / M C

Buses are assumed to have to drive more cautious than cars and trucks since passenger don’t wear

seatbelts or are standing, see Table 2 (Buses).

Table 2 Specification of the driving logics to be simulated for the three AV classes in the different road environments for the vehicle type bus.

Vehicle type: Bus



Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C

Minibuses are assumed to mainly be used for first and last mile services, hence defining driving logics

for motorways seems irrelevant. The first generation of minibuses (Basic AV) are expected to drive

according to the rail-safe driving logic on urban streets and shared spaces. The second generation

(Intermediate AV) of minibuses are also expected to be able to drive on arterials. Furthermore, the

technological development is assumed to enable more advanced driving logics so that the minibuses

3 If all conflicts between vehicles and active modes are separated, e.g. by separate phases in a traffic signal. 4 It is questionable to what extent drivers would accept the rail-safe logic driving in shared space due to the high “politeness” of the driving logic and the resulting low speed. Thus, it might from a traffic simulation point of view be more reasonable to assume manual driving


h2020-coexist.eu


Page 15 of 120

can traverse urban streets and shared space using the cautious driving logic. The Advanced AV

minibuses are expected to enable driving according to the normal driving logic in shared spaces and

urbans streets and the all-knowing driving logic on the less complex arterials.

Table 3 Specification of the driving logics to be simulated for the three AV classes in the different road environments for the vehicle type minibus.

Vehicle type: Minibus



Motorway - - -

Arterial C C / N5 AK

Urban street RS C N

Shared space RS C N

The relevant combinations of AV driving logics and functions for three typical road environments:

motorways (M), arterials (A) and urban streets (U) are specified in Table 4. The relevant combinations of

AV-driving logics and CAV-functions are described in terms of their relevance.

3.7 Relevant combinations of CAV driving logics and functions

The last step with respect to level of automation is to specify relevant combinations of CAV driving logics

and CAV-functions for different road environments. This is specified in the scenario specifications (see

Appendix C - Appendix J) by putting an M (Motorway), A (Arterial), U (Urban streets) and/or S (Shared

Space) in the relevant cells in Table 4. In cases where a more advanced version of an CAV-function is

marked less advanced function are not marked.

5 Normal driving logic might be possibly for Minibuses but not on cars due to more expensive detectors on public transport vehicles.


h2020-coexist.eu


Page 16 of 120

Table 4 Combinations of the CAV driving logics and CAV-functions. M (motorway), A (arterial), U (urban street) or S (shared space) denotes that the CAV function is included in the indicated AV-driving logic in this use case.

Driving function CAV driving logic

RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")

ACC - Adaptive Cruise Control (level 1)

LKA - Lane Keeping Assist (level 1)

PA - Park Assist (level 1)

ACC with Stop & Go (level 1)

Pedestrian Safety Systems (level 1)

Park Assistance (Level 2)

Parking Garage Pilot (Level 4)

Traffic Jam Assistance (level 2)

Traffic Jam Chauffeur (level 3)

Highway Chauffeur (level 3)

Highway Pilot (level 4)

Urban and Suburban Pilot (level 4)

Everywhere Pilot (level 5)

Co

mm

un

icati

on

fun

cti

on

s I2V (Vehicle receives info from signals)6

V2I (Signals receive information from vehicles)7

Vehicles sends/receives info to/from vehicles8

Parking space reservation & navigation9

Co

op

era

tio

n

fun

cti

on

s Merging cooperation10

Conflict resolving cooperation ("no signals needed")11

Lane change cooperation12

Platooning13

Expected as mandatory

Possible

Not expected or not possible


h2020-coexist.eu


Page 17 of 120

4 Summary of scenario specifications

4.1 Gothenburg, Sweden

4.1.1 Use case 1: Shared space (microscopic modelling)

The main uncertainty factors varied in this use case is not the penetration rate of CAVs but rather the

demand level and the adaptation of pedestrians. Demand levels include today’s demand with and

without the introduction of minibuses as well as a high pedestrian and a medium high minibus demand

level. Walking is usually the prevalent mode in shared spaces and pedestrians might adapt their

behaviour since they outnumber the rather over-cautious CAVs. The introduction of automated

minibuses in shared space are assumed only to be possible in the established and the prevalent stage.

This to be consistent with the assumption that there will be no driving logic for shared space available in

the introductory stage, see Table 1. The penetration levels and mixes of AV classes used in this use

case is summarised in Table 5.

Table 5 Penetration rate (%) and mix of AVs for different vehicle types for use case 1

Stages Vehicle type AV

penetration

AV class mix

Basic AV share

Intermediate AV share

Advanced AV share

Established Car & Truck 50 0 100 0

Minibus 100 0 100 0

Prevalent

Car & Truck 80 0 0 100

Minibus 100 0 0 100

6 A COM example made by PTV exists – vehicles adapt their speed in order to minimize the number of stops. 7 Possible using COM scripts, control strategy need to be designed. No PTV example available. 8 Possible using COM scripts, PTV example exists – vehicles with the ability to receive information adapt the speed and behavior because of an accident further downstream. 9 The reservation and navigation strategy needs to be defined. Basic functionality available in Vissim, anything beyond that would need to be programmed and might be challenging. 10 Partially available in Vissim. Anything beyond that must be programmed with COM or drivermodell.dll. 11 The control strategy needs to be defined and developed by COM. No PTV example available. Already done with Vissim by someone outside PTV (Linda Wu and Guohui Zhang, CAV Trajectory Formulation for Optimal Intersection Management and Simulation), see example at https://www.youtube.com/watch?v=4SmJP8TdWTU. 12 Partially available in Vissim. Anything beyond that must be programmed with COM or drivermodell.dll. 13 PTV example with a COM solution available. Currently under specification for development in Vissim (probably will be available as a function without the need for scripting)

https://www.youtube.com/watch?v=4SmJP8TdWTU


h2020-coexist.eu


Page 18 of 120

4.1.2 Use case 2: Accessibility during long-term construction works (macroscopic

modelling)

The main uncertainty factors varied in use case 2 is the penetration rate of CAVs for cars and trucks, see

Table 6 for an overview. The demand configuration is also varied between morning and afternoon peek

to capture potential differences, such as the congestion in the different links. In addition, the sensitivity of

the results due to behavioural adaptations of non-AV drivers in the established and prevalent stage will

be investigated. Car drivers may adapt their speed compliance, following time gap, etc. when CAV gets

established and reach higher penetration rates (e.g. larger than 50%).

Table 6 Range of penetration rates (%) and mix of AVs for different vehicle types for use case 2


penetration

AV class mix

Basic AV share


Advanced AV share

Introductory Car 10-40 80 20

Truck 10-40 80 20

Established Car 30-70 10 80 10

Truck 30-70 10 80 10

Prevalent

Car 60-90 20 80

Truck 60-90 20 80

4.2 Helmond, Netherlands

4.2.1 Use case 3: Signalised intersection including pedestrians and cyclists

(microscopic modelling)

The main uncertainty factors varied in this use case is the penetration rate of CAVs and the mix of AV

classes for cars and trucks, see Table 7 for an overview. In addition, both the current travel demand and

a future travel demand with increased car traffic are investigated.

Table 7 Range of penetration rates (%) and mix of AVs for different vehicle types for use case 3


penetration

AV class mix

Basic AV share


Advanced AV share

Introductory Car & Truck 10-40 70-100 0-30

Established Car & Truck 30-70 0-20 80-100

Prevalent Car & Truck 60-90 20-80 20-80


h2020-coexist.eu


Page 19 of 120

4.2.2 Use case 4: Transition from interurban highway to arterial (microscopic modelling)

The main uncertainty factors varied in use case 4 is the penetration rate of CAVs and the mix of AV

classes for cars and trucks, see Table 8 for an overview. In addition, both the current travel demand and

a future travel demand with increased car traffic are investigated. Furthermore, car drivers’ potential

adaptation of speed compliance, following time gap, etc. when CAV gets established and reach higher

penetration rates is also investigated.

Table 8 Range of penetration rate (%) and mix of AVs for different vehicle types for use case 4


penetration

AV class mix

Basic AV share


Advanced AV share

Introductory Car & Truck 10-40 70-100 0-30

Established Car & Truck 30-70 0-20 80-100

Prevalent Car & Truck 60-90 20-80 20-80

4.3 Milton Keynes, UK

4.3.1 Use case 5: Waiting and drop-off areas for passengers (microscopic modelling)

The penetration rate of CAVs for cars and trucks are one of the important uncertain factors, see Table 9

for an overview. Each stage of coexistence is assumed to have a clearly predominant AV class with a

small fraction of the adjacent classes. The automated buses are assumed to all be of the same class

and are assumed to not get automated until the Intermediate AV class is available (i.e. in the established

stage).

The demand levels are also considered an important uncertain factor and several different demand

levels are considered, including baseline and prognosis demand configuration which describes the case

when car access to the city centre is restricted and drop off zones has been constructed.

When CAVs become established and reach higher penetration rates, car drivers’ potential adaptation of

speed compliance, following time gap, etc. are also investigated. Pedestrians potential adaptation of

their behaviour in interactions with CAVs are also planned to be investigated together with sensitivity

analysis of boarding/alighting time at the waiting and drop off zones.


h2020-coexist.eu


Page 20 of 120

Table 9 Ranges of penetration rates (%) and mix of AVs for different vehicle types for use case 5


penetration

AV class mix

Basic AV share


Advanced AV share

Introductory Car & Truck 10-40 80 20

Bus -

Established Car & Truck 30-70 10 80 10

Bus 0, 100 100

Prevalent Car & Truck 60-90 20 80

Bus 0, 100 100

4.3.2 Use case 6: Loading and unloading areas for freight (microscopic modelling)

One of the main uncertainty factor varied in this use case is the penetration rate of CAVs for cars and

trucks, see Table 10 for an overview. Each stage of coexistence is assumed to have a clearly

predominant AV class with a small fraction of the adjacent classes. The automated buses are assumed

to all be of the same class and are assumed to not get automated until the Intermediate AV class is

available (i.e. in the established stage).

The demand levels are also considered an important uncertain factor and several different demand

levels are considered, including baseline and prognosis demand configuration which describes the case

when vehicle and delivery access to the city centre is restricted and drop off zones have been

constructed.

When CAVs become established and reach higher penetration rates, car drivers’ potential adaptation of

speed compliance, following time gap, etc. are also investigated. Pedestrians potential adaptation of

their behaviour in interactions with CAVs are also planned to be investigated together with sensitivity

analysis of boarding/alighting time of freight at the waiting and drop off zones.


h2020-coexist.eu


Page 21 of 120



penetration

AV class mix

Basic AV share


Advanced AV share


Bus -


Bus 0, 100 100


Bus 0, 100 100

4.4 Stuttgart, Germany

4.4.1 Use case 7: Impacts of CAV on travel time and mode choice on a network level

(macroscopic modelling)

The main uncertainty factors varied in use case 7 is the penetration rate of AVs for cars and trucks, see

Table 11 for an overview. It would be preferable to vary the mix of AV classes but it will be difficult for the

macroscopic use cases. The mix is assumed to be fixed in order to limit the number of microscopic traffic

simulation investigations required to derive passenger car units for the road environment categories

(motorways, arterial and urban street). The basic assumption is that, traveller’s preferences are the

same as today. However, the introduction of CAVs could imply changes in traveller’s perception for

example in terms of how they value their time. Reasons for a change in the value of time are a higher

comfort and the possibility to use the in-vehicle time more efficiently, as long as the car takes over the

control, which depends on the AV class and the road environment. Therefore, the sensitivity of the

results due to the value of time will be investigated.



penetration

AV class mix

Basic AV share


Advanced AV share





h2020-coexist.eu


Page 22 of 120

4.4.2 Use case 8: Impact of driverless car- and ridesharing services (macroscopic

modelling)

Use case 8 is to some extent different compared to the other use cases since it focuses only on the

prevalent stage with 100% driverless vehicles (cars, including ride sharing vehicles, and trucks). The

assumption to have 100% driverless vehicles within this use case implies to have a penetration rate of

100% as well as 100% of the most highly developed AV. The complexity of the use case 8 is not derived

from various mixes in penetration rate and AV fleet compositions, but from the variety of possible

mobility services and additional measures. The main uncertain factors in this use case are the

willingness of the people to share cars or rides as well as the fare model / prices for sharing services.

Table 12 Penetration rate (%) and mix of AVs for different vehicle types for use case 8


penetration

AV class mix

Basic AV share


Advanced AV share

Prevalent Car14 & Truck 100 100

5 Generic scenarios

The eight CoEXist use cases are diverse and may at first sight require individual treatment regarding the

scenarios under which to investigate them. After an initial preliminary attempt to define generic

scenarios, this was also the approach chosen by the consortium: when developing the scenarios for

each use case, the uncertainties where discussed for each use case separately, only guided by the

completely non-quantitative definition of three generic ‘stages of coexistence’. However, as the scenario

specifications neared completion, it was noted that several scenario specifications displayed remarkable

similarities in some aspects. These similarities were further increased by small modifications of some of

the individual specifications in the final stages of completion, to produce something resembling generic

scenarios.

5.1 Penetration rates

Table 13 summarize the total range of penetration levels and mix of AVs used in the different use cases

(with exception to use case 8 which describe a special 100% driverless vehicle situation at the end of the

prevalent stage). The table gives a rough definition of the different stages of coexistence in terms of

penetration rates and shares of different AV classes. The introductory stage is described by lower

penetration levels and mainly Basic AVs with potentially some Intermediate AVs. The established stage

is defined by a moderate penetration level and mainly Intermediate AVs but with the possibility with

some share of Basic and Advanced AVs. The prevalent stage is described by high penetration rates and

a wide range of shares of Intermediate and Advanced AVs.

14 Including ridesharing vehicles.


h2020-coexist.eu


Page 23 of 120

Table 13 Summary of the variation in ranges of penetration rates (%) and mix of AVs for the different vehicle types


penetration

AV class mix

Basic AV share


Advanced AV share

Introductory

Car & Truck 10-40 70-100 0-30 n.a.

Bus n.a. n.a. n.a. n.a.

Minibus n.a. n.a. n.a. n.a.

Established

Car & Truck 30-70 0-20 80-100 0-10

Bus 0,100 n.a. 100 n.a.

Minibus 100 n.a. 100 n.a.

Prevalent

Car & Truck 60-90 n.a. 20-80 20-80

Bus 0,100 n.a. n.a. 100

Minibus 100 n.a. n.a. 100

5.2 Possible behavioural adaptation

The importance of considering potential behavioural adaptation to the introduction of CAVs is deemed to

vary significantly between the use cases, see Table 14.

The possibility of adaptation of the behaviour of pedestrians are considered in use cases 1, 5, and 6.

Since use case 1 considers a shared space environment with large pedestrian flows, it is deemed

important to consider potential behavioural adaptations of the pedestrians in this use case. This is the

case also in use cases 5 and 6, where behaviour in connection to the boarding and alighting processes

are deemed uncertain and potentially important since it may strongly affect the efficiency of waiting and

drop off zones. The Helmond use cases also includes pedestrians; these are, however, separated from

motorized traffic by traffic signals, so the impacts of likely behavioural adaptations are assumed to be

small in these cases.

Adaptation of the behaviour of conventional cars as a response to the introduction of CAVs are

considered in use cases 4, 5, and 6. In use case 4 the most critical uncertainty in driver behaviour

adaptation is with respect to speed limit compliance and possibly also following time gaps, while in use

cases 5 and 6, the focus is on the behaviour at the waiting and drop off zones.


h2020-coexist.eu


Page 24 of 120

Table 14 Possible behavioural adaptations considered in the indicated use cases.

Change in interaction with vehicles in general due to CAV introduction

Conventional car drivers

Pedestrians Cyclists

Introductory 5,6 5,6

No adaptation Established 4,5,6 1,5,6

Prevalent 4,5,6 1,5,6

6 Conclusions and lessons learnt

The three types of uncertainties considered in the CoEXist scenarios (level of CAV introduction, travel

demand, and potential behavioural adaptation) do not cover all uncertainties related to the introduction of

CAVs. However, we believe that they cover the most important uncertainties that can be handled

through scenario-based analysis with the traffic simulation tools available. We would like to emphasize

the importance to include uncertainties both in the vehicle population evolution and in the behaviours of

the road users, for motorized as well as for active modes.

Just considering the uncertainties related to the vehicle fleet composition, the number of uncertain

dimensions is quite large, with general speed of the technological development within automation and

market impact of this technology being important and highly uncertain factors that often is mentioned in

discussions around the future transport system. However, when simulations of specific roads are to be

undertaken, these major uncertainties are multiplied due to the detailed assumptions needed to perform

traffic simulation experiments; it is important what automation functions are technically feasible, allowed,

and actually activated by the users, for different road environments at different stages of the introduction

of CAVs. In addition, we can also expect all these factors to be highly heterogeneous over the vehicle

population, due to various functions being available for different brands and price levels, for the different

times of production of the vehicles in the fleet.

To regard all these uncertainties as independent would be infeasible in practice due to the curse of

dimensionality; the number of simulation experiments required would become too large. We have thus

simplified the treatment of the uncertainties related to the vehicle fleet evolution by assuming that the

penetration rate of CAVs and the availability of advanced automation functions and driving logics co-

vary, and that they become available first for highly separated environments like motorways and later for

more complex environments, such as urban streets. These assumptions allow us to constrain the space

of possibilities in need of exploration to the vicinity of what we believe to be the most likely development.

Without these assumptions the problem would be intractable in practice.

A major risk with this approach is that various CAV manufacturers (or road users) may have radically

different approaches to, and principles for, connected and automated driving. One could imagine that

some manufacturers may focus completely on safety, providing only driving logics that are completely

focused on safety, similar to the rail-safe logic used in CoEXist, while other manufacturers may be

satisfied with being safer than the average human driver. In such a scenario, the rail-safe logic could


h2020-coexist.eu


Page 25 of 120

coexist with more advanced (less safe) driving logics, such as the 'normal' driving logic on the same road

types, something that would not be covered by our simplified exploration of the possible developments.

As presented in section 5, the simplified treatment of the uncertainties in the vehicle population has

allowed us to construct three generic stages of coexistence, allowing us to explore the whole range of

coexistence based in the tree stages.

Another important simplification in the treatment presented here is that we do not consider the handover

between automated and manual driving, even though we implicitly assume that such handovers are

taking place. This assumption can be thought of as assuming that relevant functions will not be available

until the handover can be handled sufficiently well to not significantly affect the simulations.


h2020-coexist.eu


Page 26 of 120

7 Partners


h2020-coexist.eu


Page 27 of 120

Appendix A Driving Logics

By Adriano Alessandrini, Markus Friedrich, and Peter Sukennik

Automated vehicles are (yet) unknown beasts (to refer to the rhinoceros’ picture). Simulating their

behaviour requires therefore many assumptions. The automated vehicles on the market today use very

simplified driving logics. The perfect technology is easier to simulate than to realise and to put on the

streets. As such, this short document proposes for the CoEXist simulation package not to embed the

driving logic of one automated-vehicle maker or the other but to use instead four exemplary logics, which

can simulate different approaches and different levels of technological development.

The first suggested logic is the one of the existing automated road vehicles, which are derived from people

movers. Such vehicles use a very simplified logic (to be certifiable with the Machinery directive) based on

the switch principle: if anything is on the collision course OFF if not ON. This can be dubbed the CityMobil2

approach or the rail-safe one. Nothing is left to chance.

The fourth and last is the “all-knowing” driving. The driver perceives everything, which is in the field of view

of the sensors and can predict always accurately the behaviour of the other road users. This automated-

driver is capable of forcing his way on other drivers whenever is needed without however ever causing

accidents.

In between the two, there are two further possible logics: the extremely cautious driving and the “normal”

driving. The extremely cautious calculates gaps accurately and only merges when gaps are acceptable

and it slows down every times its sensors can have blind angles to have no surprises. The “normal” driving

uses the logic of the average driver already modelled with the augmented (or diminished) capacities of the

sensors.

Each one of the following sections is dedicated to one of these four simplified logics. Though some details

are provided for each logic the descriptions in this document do not expect to be exhaustive but just clear

enough to start a discussion which will converge toward a sufficiently detailed definition to be embedded

in simulation software.

The results of such discussions should bring to defining formulas for gap acceptances in all conditions to

be then embedded in the simulators.

The “rail-safe” logic

To be certifiable with the machinery directive any self-moving “machine” has to have a deterministic

behaviour. This reason lead to a simple but effective (at least in the closed environments of ports and

factories) control logic for automatically moving machineries. The vehicle follows a pre-defined path for

the whole trajectory.

Should anything be in the collision course (no matter how they came to be there, their speed or trajectory),

the vehicle would brake at emergency brake deceleration. In fact, this lead to adopting some fail safe

technical solutions. Brakes are pre-actuated, as in rail rolling stocks, and the power on board is used to

de-brake (similar features applies for the steering actuator). A sudden loss of power would automatically


h2020-coexist.eu


Page 28 of 120

activate the brakes. This is a safe manoeuvre only up to a certain speed and depending on the restraining

systems for the people on-board. As such when brought on the streets the speed of these systems was

limited (40km/h so far) and they were complemented with additional obstacle detection features.

The vehicle maximum velocity is set at design stage depending on the distance from fixed obstacles, which

can hide threats and in operation the actual speed is set after calculating time to collision as shown in

Figure 2 below.

Figure 2: Illustration of the time to collision calculation

Time to collision Tc is

And the vehicle velocity is calculated as

with

• a = vehicle acceleration (negative when braking)

• 𝑡𝑅 = A reaction time

The obstacle detection sensors draw around the vehicle two “dumb” safety areas; the collision course,

which is laterally delimited by the width of the vehicle itself and has a length which depends on the speed

and the emergency deceleration of the vehicle. Anything in this area triggers the emergency stop. The

second area is not only in front of the vehicle but on its back and sides too and speed reductions are

mandated if any object is identified in such area with the potential to enter the collision course. The

maximum vehicle speed is decided section by section, while certifying the infrastructures, depending on

the fixed obstacles, which might hide moving ones. The real speed of the vehicle will be at best the

maximum certified speed for the trajectory if there are no moving obstacles in sight otherwise it will be

lowered in case a moving obstacle (even outside its immediate collision course) can cause a potential

threat. A pedestrian on the sidewalk, even if detected when walking in parallel to the vehicle trajectory,

needs to be tracked and the vehicle speed must be adapted to be ready to brake in case she suddenly

changes courses and jumps in front of the automated vehicle.

T𝐶 = 𝑡𝑅 −𝑉𝐴𝑅𝑎

𝑉𝐴𝑅 = 𝑎 ∙ 𝑡𝑅 −𝑎 ∙ 𝑑𝐵𝑉𝐵


h2020-coexist.eu


Page 29 of 120

The example of Figure 3 below is an urban motorway with three lanes per direction, un-signalised

intersections at grade and unprotected sidewalks. An infrastructure as such will never be certified for the

“rail-safe” logic but nevertheless is a good example to show what can and cannot be done with this logic

and the others.

In the left image of Figure 3 the automated vehicle is all alone. It has a full view on the sidewalk, which is

empty and can therefore ride at its maximum speed. As soon as a pedestrian is present on the sidewalk

the speed of the vehicle is limited by the time to collision with such “obstacle” should she decide to suddenly

cross the street. If the sidewalk is a metre away from the automated vehicle course and a pedestrian walks

at a meter per second, the vehicle can only overtake the pedestrian at a speed which would allow it a full

stop in one second. If the maximum allowed deceleration is 1.2 meters per square seconds (standing non-

restrained passengers on board) the maximum speed can be 1.2 meters per second (nearly 4.5 km/h);

which grows to 18 km/h with sitting and buckled passengers on board. To increase the vehicle speed the

sidewalk needs either to be segregated or some safety boundary put between the sidewalk and the

automated vehicle lane.

As shown in the second image in Figure 3 the vehicle cannot change lane and, as shown in the third, it

cannot turn left. To turn left the vehicle must have a clearly marked lane (as shown in the fourth image)

and a dedicated traffic-light aspect when all other vehicles stop and the automated vehicle passes.

Figure 3 “Rail safe” logic for road following (left-hand side) and prohibited lane changing (second image from the left) and left turning (third image from the left). To allow left turning for a “rail safe” logic the road need to be clearly marked, a traffic light inserted with a dedicated aspect for the automated vehicle left turning which can even turn left from the right lane at this point (right-hand side).

The vehicle in this logic can have no deviations from a pre-certified trajectory and the only emergency

manoeuvre would be braking.

To coexist with the other vehicles the distance from the vehicle in front will be calculated with the Brick-

Wall-Stop criterion and even following vehicles would need to follow at such distance. In case they do not

respect it, the vehicle would reduce speed to allow smaller distances to be safe.

The automated vehicle would check on the right-hand side for pedestrians and eventually adapt speed as

described above while overtaking cars on the middle lanes would be difficult to handle. In fact, this

infrastructure, to be certifiable with the “rail safe” logic, would need some kind of buffer between one lane

and the other; either some gap or a form of segregation.


h2020-coexist.eu


Page 30 of 120

The “cautious driving” logic

When originally brought to the streets the Google Cars slowed and stopped for any potential danger thus

happening to be often rear-ended. The car was then trained to become more aggressive. However, more

aggressive also means less safe. This choice brought to the first accidents in which the Google Car was

at fault according to the road code. The accident did not happen because any sensor or system failure of

any kind but because the car assumed that moving on the left in front of a bus, which had priority, it would

have slowed down yielding and it did not. For the google car it was either this, forcing its way, or stopping

in the middle of the road and waiting for a moment which might never come. Drivers do so every day (in

fact they do have accidents) but they use eye contact and other forms of communications to be “safe”.

The cautious driving logic (though either abandoned or soon to be by car makers) is extremely useful

because it gives the measure of which the performance of the transport system would be adopting a

behaviour in which the automated vehicle is never the one responsible for an accident. This does not mean

(as for the rail safe) not being involved in accidents because other road users can still cause them and

involve the automated vehicle; it means only never being at fault for one. While the rail safe imposes to

slow down if a pedestrian is on the sidewalk even if she is minding her own business and could be ignored,

the cautious driver would ignore her and in case she jumps on the hood she dies but it will be her fault.

She will still be dead but the “cautious driver” would bear no responsibility for it; and, knowing machines,

it would not feel guilty.

It is (relatively) easy to implement this logic; the vehicle respects the road-code and the “safe behaviour”

always. It calculates the gap with the other moving obstacles and if there is no completely safe passage it

waits slowing and eventually stopping. If the sensors have not a clear view (there is a car or any other

obstacle impeding them) they assume there is an obstacle behind and the vehicle behaves safely.

For example, car-following will only happen at BWS (Brick Wall Stop) distance (as shown in the first image

of Figure 4 below).

Any manoeuvre needs to allow the vehicles on the conflicting paths not to brake until the manoeuvre is

completed and even after the completion of the manoeuvre a minimum brick-wall distance needs still to

be in place.

In the third image of Figure 4 the vehicle is allowed to turn left. However, it can do so only if each vehicle

in the incoming lane is far enough not to decelerate. Assuming each vehicle will continue with its own

speed and given the speed of the automated vehicle when the automated vehicle clears the path of the

incoming vehicle a BWS distance needs still to be in place. Should for any reason the automated vehicle

stop the incoming vehicle can still brake and not hit the vehicle but in case the automated vehicle does not

stop the incoming vehicle does not decelerate.

Similarly for the lane changing (centre image in Figure 4) the lane changing manoeuvre is considered

completed when the automated vehicle has reached the same speed of the flow of vehicles in the lane

and the vehicle in the back (at the same speed) still has a BWS distance from the automated vehicles.

This leads to difficult lane change in heavy traffic.


h2020-coexist.eu


Page 31 of 120

Figure 4: Gap acceptance for car-following, lane changing and left turning in the “cautious driving” logic

The “normal driving” logic

This logic is more a benchmark than anything else. It uses the existing average driver already modelled

and gives him the capacity of measuring the distances and speeds of the other vehicles the sensor suit

will give. Sensor sight and occlusion might be relevant to simulate what the “normal driver” controlled

vehicle knows for real.

This is probably the most difficult logic to implement for real but the easiest to model.

The “all-knowing driving” logic

As explained when discussing the cautious driving logic, unless the automated vehicles will have simplified

roads they will need to mimic the human driver behaviour and its aggressive behaviour. It is often needed

to force own way when changing lane, turning left, or making any other manoeuvre which breaks an

incoming flow of cars. But it is necessary to negotiate its own way with cyclists, pedestrians and all other

road users. This is the reason why all the players researching control logic for automated vehicles are now

entering artificial intelligence realm to predict other road user behaviours. This is still far to come and many

hopes that machine learning techniques will help teaching computers to predict such behaviours and

automatically adapt to it. CoEXist will not simulate machine learning processes, it will just assume that the

outcomes of the learning process is a driving logic which (together where necessary with V2X

communication) allows the vehicle to be all-knowing and to exploit this knowledge to have the best possible

performances while respecting safety.

Such driving logic is much easier to simulate rather than implementing it for real. All the objects in the

simulations have known positions, speeds and behaviours. Assuming the object is in view of the sensors

(or is hidden, but has been recently in view or in view to a connected vehicle, which passes on the

information) this control logic can trace the obstacle and perfectly predict their future behaviour including

the reaction to “barging” in their trajectories and adapt own behaviour to the predicted reaction which will

be exactly the foreseen one.


h2020-coexist.eu


Page 32 of 120

This will also go for car-following models. Being the automated vehicle capable of predicting any sudden

brake of the vehicle in front the gap to keep can become rather smaller.

Figure 5: Gap acceptance for car-following, lane changing and left turning in the “all-knowing driving” logic

Lane changing at the centre of Figure 5 happens with such small gaps that the following vehicle yield to

let it merge as in a sort of cooperative manoeuvre.

Left turning on the right of Figure 5 can be done with small gaps and even when some of the vehicles in

the furthest lanes are not in sight.

Such behaviour can either happen because prediction models are perfect or because there is cooperation between vehicles however in simulation it is not important which is the technology but only the resulting behaviour.


h2020-coexist.eu


Page 33 of 120

Appendix B Descriptions of AV-functions

By Peter Sukennik

This appendix includes a list of AV-functions with a brief description as stated in the document

“Deployment paths for Vehicle and Road Automation (Final)” (D3.1.3), created under the project

“Support action for Vehicle and Road Automation network”.


The cruise control system with “Adaptive distance control ACC” uses a distance sensor to measure

the distance and speed relative to vehicles driving ahead, usually using perception information

coming from cameras and lasers. The driver sets the speed and the required time gap.

LKA Lane Keeping Assist (level 1)

LKA automatically becomes active from a specific speed (normally from 40 mph) and upwards. The

system detects the lane markings and works out the position of the vehicle. If the car starts to drift off

lane, the LKA takes corrective action. If the maximum action it can take is not enough to stay in lane,

or the speed falls below 40 mph, LKA function warns the driver (e.g. with a vibration of the steering

wheel). Then it's up to the driver to take correcting action.


PA automatically steers the car into parallel and bay parking spaces, and also out of parallel parking

spaces. The system assists the driver by automatically carrying out the optimum steering movements

to reverse-park on the ideal line. The measurement of the parking space, the allocation of the

starting position and the steering movements are automatically undertaken by the PA – all the driver

has to do is operate the accelerator and the brake. This means that the driver retains control of the

car at all times.

ACC with Stop&Go (level 1)

Adaptive Cruise Control with stop & go function includes automatic distance control (control range

30–250 km/h) and, within the limits of the system, detects a preceding vehicle. It maintains a safe

distance by automatically applying the brakes and accelerating. In slow-moving traffic and

congestion, it governs braking and acceleration.


Pedestrian detection systems are mostly used for urban environment. The pedestrian detection can

be classified like a Collision Warning System (CWS, Level 0). However, since the reaction time of the

driver is slow (around 2 seconds), these systems usually have access to the brake system

(longitudinal control).


Partial Automated Parking into and out of a parking space, working on public parking area or in

private garage. Via smartphone or key, parking process is started and vehicle accomplishes parking

maneuvers by itself. The driver can be located outside of the vehicle, but has to constantly monitor

the system, stops parking maneuverer if required


Highly Automated parking includes maneuvering to and from parking place (driverless valet parking).

In parking garage, the driver does not have to monitor the system constantly and may leave once the

system is active. Via smartphone or key parking maneuverer and return of the vehicle is initiated.


h2020-coexist.eu


Page 34 of 120


The function controls the vehicle longitudinal and lateral to follow the traffic flow in low speeds

(<30km). The system can be seen as an extension of the ACC with Stop&Go functionality.


Conditional automated driving in traffic jam up to 60 km/h on motorways and motorway similar roads.

The system can be activated, if traffic jam scenario exists. It detects slow driving vehicle in front and

then handles the vehicle both longitudinal and lateral. Driver must deliberately activate the system,

but does not have to monitor the system constantly. Driver can at all times override or switch off the

system. Note: There is no take over request to the driver from the system.


Conditional Automated Driving up to 130 km/h on motorways or motorway similar roads. From

entrance to exit, on all lanes, incl. overtaking. The driver must deliberately activate the system, but

does not have to monitor the system constantly. The driver can at all times override or switch off the

system. The system can request the driver to take over within a specific time, if automation gets to its

system limits.


Automated Driving up to 130 km/h on motorways or motorway similar roads from entrance to exit, on

all lanes, incl. overtaking. The driver must deliberately activate the system, but does not have to

monitor the system constantly. The driver can at all times override or switch off the system. There are

no requests from the system to the driver to take over when the systems in normal operation area

(i.e. on the motorway). Depending on the deployment of cooperative systems ad-hoc convoys could

also be created if V2V communication is available.


Highly automated driving up to limitation speed, in urban and suburban areas. The system can be

activated by the driver on defined road segments, in all traffic conditions, without lane change in the

first phase. The driver can at all-time override or switch off the system. This system handles very

dynamic scenarios, including: pedestrian, motorcycles, bikes, etc.

Everywhere pilot (level 5)

The everywhere pilot should be able to handle all driving from point A to B, without any input from the

passenger. The driver can at all-time override or switch off the system. Note: no realistic time

estimation exists on this system.


h2020-coexist.eu


Page 35 of 120

Appendix C Use case 1: Shared space

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No. 723201-2

Scenario specification

for use case 1

Shared space

Version: 1.0

Date: 2018-03-20

Author: Fredrik Johansson and Johan Olstam





h2020-coexist.eu


Page 37 of 120

Table of contents

1 Level of automation ................................................................................... 38

1.1 Relevant combinations of CAV logics and functions .......................................................... 40

2 Traffic demand ........................................................................................... 42

2.1 Traffic demand configurations ........................................................................................... 42

2.1.1 Demand configuration 1.......................................................................................................... 42




2.2 Relevant demand configurations per stage of coexistence ............................................... 45

3 Road user behaviour for non AVs ............................................................ 46


h2020-coexist.eu


Page 38 of 120

1 Level of automation

Table 15 presents indicative penetration rates of AVs for cars and trucks. Car and truck volumes are low

and will constitute background traffic in this use case, so no specific investigation into the sensitivity to

AV penetration rates for cars is deemed needed.

Table 16 presents the AV penetration rate for pods. These are assumed to be introduced at a stage

when vehicles are able to drive without driver through the city centre.

Table 15 Penetration rates of AV of vehicle types cars and trucks for each stage of coexistence.

Vehicle type: Cars & trucks

Stage AV penetration rate

Introductory

Established 50

Prevalent 80

Table 16 Penetration rates of AV of vehicle type pod for each stage of coexistence.

Vehicle type: Pods


Introductory

Established 100

Prevalent 100

Table 17 present the composition of AVs in terms of the AV-classes: Basic, Intermediate, Advanced

(specified in Table 1 - Table 3). All vehicle types are assumed to follow the same development trajectory,

here simplified to one AV class per stage.


h2020-coexist.eu


Page 39 of 120

Table 17 Indicative ratios of AV for each AV class for each stage of coexistence.

Vehicle type: Cars, Trucks and Pods

Stage Ratios of AV class

Basic AV Intermediate

AV Advanced

AV

Introductory

Established 100

Prevalent 100

It is assumed that it is technically feasible for cars with intermediate AV to drive with RS driving logic in

shared spaces. The penetration rate for intermediate AVs in shared space environments is in this use

case interpreted as the fraction that actually use it.

Table 18 Specification of which driving logic that Basic AV, Intermediate AV and Advanced AV cars will utilize in different types of road environments.

Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M RS C


h2020-coexist.eu


Page 40 of 120

Table 19 Specification of which driving logic that Basic AV, Intermediate AV and Advanced AV trucks will utilize in different types of road environments.

Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C

Table 20 Specification of which driving logic that Basic AV, Intermediate AV and Advanced AV pods will utilize in different types of road environments.

Vehicle type: Pod



AV Advanced AV

Motorway - - -

Arterial - N AK

Urban street RS C N

Shared space RS C N

1.1 Relevant combinations of CAV logics and functions

The relevant combinations of AV driving logics and functions are specified in Table 21.

All vehicles are assumed to be able to drive automated through a shared space (more or less efficiently);

no additional functions will be required. The pedestrian safety systems may require extra attention since

these may influence the performance of traffic through shared spaces.


h2020-coexist.eu


Page 41 of 120

Table 21 Combinations of AV driving logics and CAV-functions. M (Motorway), A (Arterial), U (Urban street) or S (Shared Space) denotes that the CAV function is included in the indicated AV-driving logic in this use case.

Driving function AV driving logic

RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")





Pedestrian Safety Systems (level 1) US US US







Urban and Suburban Pilot (level 4) US US


Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning22

Expected as obligatory

Possible



h2020-coexist.eu


Page 42 of 120

2 Traffic demand

2.1 Traffic demand configurations

2.1.1 Demand configuration 1

The demand configuration describes the current traffic: a shared space dominated by the pedestrian

flows. This demand configuration will only be used for calibration.

Indicative traffic demand levels per mode are presented in Table 22.

Table 22 Indicative traffic demand levels per mode.

Demand level Low Medium High

Cars X

Trucks X

Pedestrians X

Bikes X

How route choice for the relevant modes will be handled in the simulation is presented in Table 23.




h2020-coexist.eu


Page 43 of 120

Table 23 Description of how route choice will be handled in the demand configuration.

Route choice Varied Static as today Static alternative route choice

Cars X

Trucks X

Pedestrians X

Bikes X


The demand configuration describes a future traffic condition identical to the current traffic, but with the

addition of a last mile pod service with a low demand level.




Cars X

Trucks X

Pedestrians X

Bikes X

Pods X




Cars X

Trucks X

Pedestrians X

Bikes X


h2020-coexist.eu


Page 44 of 120

Pods X


The demand configuration describes a future traffic condition identical to the current traffic, but with the

addition of a last mile pod service with a medium demand level.




Cars X

Trucks X

Pedestrians X

Bikes X

Pods X




Cars X

Trucks X

Pedestrians X

Bikes X

Pods X


The demand configuration describes a future traffic condition when the city has succeeded with its

ambition to increase walking and biking, and a last mile pod service has been introduced.



h2020-coexist.eu


Page 45 of 120



Cars X

Trucks X

Pedestrians X

Bikes X

Pods X




Cars X

Trucks X

Pedestrians X

Bikes X

Pods X

2.2 Relevant demand configurations per stage of coexistence

Table 30 describe the demand levels (according to the definitions in section 3.1) that will be investigated

for each stage of coexistence. All the future demands will be simulated for both the established and

prevalent stages of automation.

Table 30 Relevant demand configurations for each stage of coexistence.

Demand level Demand 1 Demand 2 Demand 3 Demand 4

Introductory

Established X X X

Prevalent X X X


h2020-coexist.eu


Page 46 of 120

3 Road user behaviour for non AVs

The basic assumption is that non CAVs (vehicles, pedestrians, bicyclists, etc.) interact with CAVs in the

same way as they interact with manually driven vehicles today. However, the introduction of CAVs could

imply changed behaviour interaction with vehicles in general. Table 31 describe the potential road use

behaviour adaptation that will be considered for this use case.

Pedestrians are the dominant mode and may be more likely to adapt their behaviour since they may be

empowered by their numerical dominance and the caution of the CAVs.

Table 31 Road user behaviour adaptation due to the introduction of CAVs for each stage of coexistence.



Pedestrians Bicyclists

Introductory

Established X

Prevalent X


h2020-coexist.eu


Page 47 of 120

Appendix D Use case 2: Accessibility during long-term construction works


Version: 1.0

Date: 2018-04-04

Author: Johan Olstam, Nina Galligani and Fredrik

Johansson


for use case 2

Accessibility during long-term

construction works





h2020-coexist.eu


Page 49 of 120

Table of contents


1.1 Relevant combinations of CAV driving logics and functions .............................................. 51

2 Traffic demand ........................................................................................... 54




2.2 Relevant traffic demand configuration per stages of coexistence ...................................... 55

3 Traveller adaptation ................................................................................... 55


h2020-coexist.eu


Page 50 of 120


Table 15 present indicative penetration rates of AVs for cars and trucks, which is the two vehicle types

included in the use case. The main uncertain factor in this use case is the penetration rate of various

CAVs. Cars and trucks are assumed to have the same AV penetration rate.

Table 32 Penetration rates of AV of vehicle type car & trucks for each stage of coexistence.

Vehicle type: Car & Trucks


Introductory 10-40

Established 30-70

Prevalent 60-90


(specified in Table 1 - Table 35). It would be preferable to vary the mix of AV classes but it will be difficult

for the macroscopic use cases. The mix is assumed to be fix in order to limit the number of microscopic

traffic simulation investigations required to derive passenger car units for the road environment

categories (motorways, arterial and urban street).

Table 33 Indicative ratios of AV of vehicle type car for each AV class for each stage of coexistence.

Vehicle type: Car & Trucks



AV Advanced

AV

Introductory 80 20

Established 10 80 10

Prevalent 20 80

Table 1 - Table 35 present a suggestion of according to which AV driving logic (rail safe, cautions,

normal, all-knowing) that the different AV-classes (basic, intermediate, advanced) are estimated to drive

according to.


h2020-coexist.eu


Page 51 of 120


Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C


Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C



motorways (M), arterials (A) and urban streets (U) are specified in Table 4. The relevant combinations of

AV-driving logics and CAV-functions are described in terms of their relevance. Functions relevant for

shared space is not presented since the macroscopic model do not include shared space road

environments.

It is expected that a vehicle of the first generation of AV (Basic AV) will be able to handle situations on

motorways and on arterials on its own with high caution as well as the option to fall back to the manual

driver. For intermediate AV this capability will be extended to urban road environments. Additionally, the

included highway pilot doesn't need a driver as a fall back on motorways anymore. Finally, the advanced


h2020-coexist.eu


Page 52 of 120

AV is capable to handle driving in all three road environments on its own, yet it is still not able to manage

all situations under all circumstances, which is why it is not a driverless vehicle


h2020-coexist.eu


Page 53 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")








Traffic Jam Assistance (level 2) U

Traffic Jam Chauffeur (level 3) A

Highway Chauffeur (level 3) M

Highway Pilot (level 4) M A / M

Urban and Suburban Pilot (level 4) U / A U / A


Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning30


Possible



h2020-coexist.eu


Page 54 of 120

2 Traffic demand



The demand configuration describes the current travel demand situation (base year is 2017), but with

construction situation of the network corresponding to year 2019.

Time period: Morning peak




Cars X

Trucks X


The demand configuration describes the current travel demand situation (base year is 2017), but with

construction situation of the network corresponding to year 2019.

Time period: Afternoon peak


23 A COM example made by PTV exists – vehicles adapt their speed in order to minimize the number of stops. 24 Possible using COM scripts, control strategy need to be designed. No PTV example available. 25 Possible using COM scripts, PTV example exists – vehicles with the ability to receive information adapt the speed and behavior because of an accident further downstream. 26 The reservation and navigation strategy needs to be defined. Basic functionality available in Vissim, anything beyond that would need to be programmed and might be challenging. 27 Partially available in Vissim. Anything beyond that must be programmed with COM or drivermodell.dll. 28 The control strategy needs to be defined and developed by COM. No PTV example available. Already done with Vissim by someone outside PTV (Linda Wu and Guohui Zhang, CAV Trajectory Formulation for Optimal Intersection Management and Simulation), see example at https://www.youtube.com/watch?v=4SmJP8TdWTU. 29 Partially available in Vissim. Anything beyond that must be programmed with COM or drivermodell.dll. 30 PTV example with a COM solution available. Currently under specification for development in Vissim (probably will be available as a function without the need for scripting).



h2020-coexist.eu


Page 55 of 120



Cars X

Trucks X

2.2 Relevant traffic demand configuration per stages of coexistence


for each stage of coexistence.


Demand level Demand 1 Demand 2

Introductory X X

Established X X

Prevalent X X

3 Traveller adaptation

The basic assumption is that traveller’s preferences is the same as today. However, the introduction of

CAVs could imply changes in traveller’s preferences for example in terms of how the value their time.

Table 31 describe the potential changes in traveller’s preferences that will be considered for this use

case.

Table 40 Traveller preference adaptation due to the introduction of CAVs for each stage of coexistence.

Change in traveller behaviour

Value of time Other (specify)

Introductory

Established

Prevalent


h2020-coexist.eu


Page 56 of 120



imply changed behaviour interaction with vehicles in general. Table 41 describe the potential road use


Car drivers may adapt speed compliance, following time gap, etc. when CAV gets established and reach

higher penetration rates (e.g. larger than 50%).

Table 41 Road user behaviour adaptation due to the introduction of CAVs for each stage of coexistence. M (Motorway), A (Arterial) or U (Urban street) denotes for which road categories that the potential behaviour adaptation is relevant for.




Introductory

Established X

Prevalent X


h2020-coexist.eu


Page 57 of 120

Appendix E Use case 3: Signalised intersection including pedestrians and cyclists



for use case 3

Signalised intersection including

pedestrians and cyclists

Version: 1.0

Date: 2018-03-20






h2020-coexist.eu


Page 59 of 120

Table of contents



2 Traffic demand ........................................................................................... 63




2.2 Relevant traffic demand configurations per stage of coexistence ...................................... 64



h2020-coexist.eu


Page 60 of 120


Table 15 present indicative penetration rates of AVs for cars and trucks. The main uncertain factor in this

use case is the penetration rate of various CAVs. Overlaps between the penetration rates for the stages

are needed to compare the different AV classes.

Table 42 Penetration rates of AV for each stage of coexistence.

Vehicle type: Cars & Trucks


Introductory 10-40

Established 30-70

Prevalent 60-90


(specified in Table 1 - Table 45). It is assumed that cars and trucks follow the same development in

terms of AV class, and that there will be a mix of different AV classes present a given moment. The mix

is simplified to only include two AV classes simultaneously to limit the number of scenarios.





AV Advanced

AV

Introductory 70-100 0-30

Established 0-20 80-100

Prevalent 20-80 20-80

Since all conflicts with vulnerable road users are time separated in this use case, it is reasonable to

expect the normal driving logic for the Intermediate AV class instead of the cautious driving logic.


h2020-coexist.eu


Page 61 of 120


Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C N AK

Urban street M C N

Shared space M M C


Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C N AK

Urban street M C N

Shared space M M C



The cautious driving logic is expected to provide automation with lane change only at lower speeds,

while the more intermediate logics should be able to operate in the whole study site without input from

the driver, and possibly get, and act on information from the traffic lights, enabling slowing down to reach

the intersection in time for the green phase.


h2020-coexist.eu


Page 62 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")




ACC with Stop & Go (level 1) A








Urban and Suburban Pilot (level 4) A A


Co

mm

un

icati

on

fun

cti

on

s I2V (Vehicle receives info from signals)31 A A

V2I (Signals receive information from vehicles)32 A A

Vehicles sends/receives info to/from vehicles33 A A


Co

op

era

tio

n

fun

cti

on




Platooning38


Possible



h2020-coexist.eu


Page 63 of 120

2 Traffic demand



The demand configuration describes the current traffic.




Cars X

Trucks X

Pedestrians X

Bikes X





h2020-coexist.eu


Page 64 of 120



Cars X

Trucks X

Pedestrians X

Bikes X


The demand configuration describes a future scenario with large car demand.

Indicative traffic demand levels per mode is presented in Table 22.



Cars X

Trucks X

Pedestrians X

Bikes X




Cars X

Trucks X

Pedestrians X

Bikes X

2.2 Relevant traffic demand configurations per stage of coexistence


h2020-coexist.eu


Page 65 of 120


for each stage of coexistence. Both the higher and lower demands will be tested in the established and

prevalent stages; however, the introductory phase is to close in time for a large increase in demand to be

realistic.



Introductory X

Established X X

Prevalent X X




imply changed interaction with vehicles in general. Table 31 describe the potential road use behaviour

adaptation that will be considered for this use case.

No adaptation of other road users is expected or will at least not be considered in this use case.





Introductory

Established

Prevalent


h2020-coexist.eu


Page 66 of 120

Appendix F Use case 4: Transition from interurban highway to arterial



for use case 4

Transition from interurban highway to

arterial

Version: 1.0

Date: 2018-03-20






h2020-coexist.eu


Page 68 of 120

Table of contents



2 Traffic demand ........................................................................................... 72




2.2 Relevant traffic demand configurations per stage of coexistence ...................................... 73



h2020-coexist.eu


Page 69 of 120


Table 15 present indicative penetration rates of AVs for cars and trucks. The main uncertain factor in this

use case is the penetration rate of various CAVs. Overlaps between the penetration rates for the stages

are needed to compare the different AV classes.

Table 53 Penetration rates of AV of vehicle type car and truck for each stage of coexistence.



Introductory 10-40

Established 30-70

Prevalent 60-90


(specified in Table 1 - Table 56). It is assumed that cars and trucks follow the same development in

terms of AV class, and that there will be a mix of different AV classes present a given moment. The mix

is simplified to only include two AV classes simultaneously to limit the number of scenarios.





AV Advanced

AV

Introductory 70-100 0-30

Established 0-20 80-100

Prevalent 20-80 20-80

Since all conflicts with vulnerable road users are time separated in this use case, it is reasonable to

expect the normal driving logic for the Intermediate AV class instead of the cautious driving logic.


h2020-coexist.eu


Page 70 of 120


Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C N AK

Urban street M C N

Shared space M M C


Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C N AK

Urban street M C N

Shared space M M C



The cautious driving logic is expected to provide automation with lane change only at lower speeds,

while the more intermediate logics should be able to operate in the whole study site without input from

the driver, and possibly get, and act on information from the traffic lights, enabling slowing down to reach

the intersection in time for the green phase.


h2020-coexist.eu


Page 71 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")




ACC with Stop & Go (level 1) MA





Traffic Jam Chauffeur (level 3) MA


Highway Pilot (level 4) M M

Urban and Suburban Pilot (level 4) A A


Co

mm

un

icati

on

fun

cti

on

s I2V (Vehicle receives info from signals)39 MA MA

V2I (Signals receive information from vehicles)40 MA MA

Vehicles sends/receives info to/from vehicles41 MA MA


Co

op

era

tio

n

fun

cti

on




Platooning46


Possible



h2020-coexist.eu


Page 72 of 120

2 Traffic demand



The demand configuration describes the current traffic.




Cars X

Trucks X

Pedestrians X

Bikes X





h2020-coexist.eu


Page 73 of 120



Cars X

Trucks X

Pedestrians X

Bikes X


The demand configuration describes a future scenario with large car demand.

Indicative traffic demand levels per mode is presented in Table 22.



Cars X

Trucks X

Pedestrians X

Bikes X




Cars X

Trucks X

Pedestrians X

Bikes X

2.2 Relevant traffic demand configurations per stage of coexistence


h2020-coexist.eu


Page 74 of 120


for each stage of coexistence. Both the higher and lower demands will be tested in the established and

prevalent stages; however, the introductory phase is to close in time for a large increase in demand to be

realistic.



Introductory X

Established X X

Prevalent X X






Car drivers may adapt speed compliance, following time gap, etc. when CAV gets established and reach

higher penetration rates (e.g. larger than 50%).





Introductory

Established X

Prevalent X


h2020-coexist.eu


Page 75 of 120

Appendix G Use case 5: Waiting and drop-off areas for passengers



for use case 5

Waiting and drop-off areas for

passengers

Version: 0.9

Date: 2018-03-20

Authors: Brian Matthews, Ammar Anwar and

Prof John Miles





h2020-coexist.eu


Page 77 of 120

Table of contents



2 Traffic demand ........................................................................................... 82










h2020-coexist.eu


Page 78 of 120


Table 15 present indicative penetration rates of AVs for cars. The use case is in general very ambitious,

so it is important to keep the number of scenarios as low as possible. Still the whole range of penetration

rates need to be covered, and since the stages of coexistence contains different mixes of automated

vehicles, some overlap of the penetration rates of the stages may be useful. Cars and trucks are

assumed to have the same AV penetration rate, while the buses are assumed to all either be manually

driven or all automated, to reduce the number of scenarios.

Table 64 Penetration rates of AV of vehicle type car for each stage of coexistence.



Introductory 10-40

Established 30-70

Prevalent 60-90


Vehicle type: Bus


Introductory -

Established 0, 100

Prevalent 0, 100


(specified in Table 1 - Table 2). The ratios of AV classes are kept as simple as possible with no variation,

again to reduce the number of scenarios. Each stage of coexistence is assumed to have a clearly

predominant AV class with a small fraction of the adjacent classes. The automated buses are assumed

to all be of the same class, and are assumed to not get automated until the intermediate AV class is



h2020-coexist.eu


Page 79 of 120





AV Advanced

AV

Introductory 80 20


Prevalent 20 80


Vehicle type: Bus



AV Advanced

AV

Introductory - - -

Established - 100 -

Prevalent - - 100


Vehicle type: Car



AV Advanced AV

Arterial C C AK

Urban street M C N


h2020-coexist.eu


Page 80 of 120


Vehicle type: Truck



AV Advanced AV

Arterial C C AK

Urban street M C N


Vehicle type: Bus



AV Advanced AV

Arterial C C AK

Urban street M C N



The road types in Milton Keynes that would be considered in this use case are arterials and urban

streets. Most roads external to the centre and near the vehicle intercept locations on the edge of the

centre would be classified as Arterial. Within the vehicle intercept locations and within the centre roads

are considered to be urban street in behaviour.


h2020-coexist.eu


Page 81 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")







Parking Garage Pilot (Level 4) U


Traffic Jam Chauffeur (level 3) A U



Urban and Suburban Pilot (level 4) A U

Everywhere Pilot (level 5) A

Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning54


Possible



h2020-coexist.eu


Page 82 of 120

2 Traffic demand



The demand configuration describes the current demand. This demand configuration will only be used

for calibration and comparison.




Cars X

Trucks X

Buses X





h2020-coexist.eu


Page 83 of 120



Cars X

Trucks X

Buses X


The demand configuration describes the demand for testing various detailed designs of vehicle intercept

locations.




Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


h2020-coexist.eu


Page 84 of 120



locations and their bottleneck capacities. This demand will help in testing various detailed designs for the

vehicle intercept locations further.




Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


The demand configuration describes a prognosis in the case when car access to the city centre is

restricted, vehicle intercept locations constructed, and services from vehicle intercept locations into the

centre provided. This will be testing the network effects with the introduction of these measures.



h2020-coexist.eu


Page 85 of 120



Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


The demand configuration describes a prognosis in the case when car access to the city centre is

restricted, vehicle intercept locations constructed, and services from vehicle intercept locations into the

centre provided. This will be testing the effect of the final measures any further restriction changes

deemed necessary will be implemented.




Cars X

Trucks X

Buses X



h2020-coexist.eu


Page 86 of 120



Cars X

Trucks X

Buses X



for each stage of coexistence. The case with and without restriction of car traffic to the centre are studied



Demand level Demand 1 Demand 2 Demand 3 Demand 4 Demand 5

Introductory X X X X

Established X X X X

Prevalent X X X X




imply change behaviour interaction with vehicles in general. Table 31 describe the potential road use


Sensitivity analysis would be undertaken as part of the use case to investigate the distribution of people

in private cars. Also during the designing of vehicle intercept locations sensitivity analysis will be

undertaken to asses boarding/alighting time and its variance with different boarding/alighting strategies

such as boarding sides and gaps between vehicles.


h2020-coexist.eu


Page 87 of 120





Introductory X X

Established X X

Prevalent X X


h2020-coexist.eu


Page 88 of 120

Appendix H Use case 6: Loading and unloading areas for freight



for use case 6

Loading and unloading areas for

freight

Version: 0.9

Date: 2018-03-20

Authors: Brian Matthews, Ammar Anwar and Prof

John Miles





h2020-coexist.eu


Page 90 of 120

Table of contents



2 Traffic demand ........................................................................................... 95










h2020-coexist.eu


Page 91 of 120


Table 15 present indicative penetration rates of AVs for cars. The use case is in general very ambitious,

so it is important to keep the number of scenarios as low as possible. Still the whole range of penetration

rates need to be covered, and since the stages of coexistence contains different mixes of automated

vehicles, some overlap of the penetration rates of the stages may be useful. Cars and trucks are

assumed to have the same AV penetration rate, while the buses are assumed to all either be manually

driven or all automated, to reduce the number of scenarios.




Introductory 10-40

Established 30-70

Prevalent 60-90


Vehicle type: Bus


Introductory -

Established 0, 100

Prevalent 0, 100


(specified in Table 1 - Table 2). The ratios of AV classes are kept as simple as possible with no variation,

again to reduce the number of scenarios. Each stage of coexistence is assumed to have a clearly

predominant AV class with a small fraction of the adjacent classes. The automated busses are assumed

to all be of the same class, and are assumed to not get automated until the intermediate AV class is



h2020-coexist.eu


Page 92 of 120





AV Advanced

AV

Introductory 80 20


Prevalent 20 80


Vehicle type: Bus



AV Advanced

AV

Introductory - - -

Established - 100 -

Prevalent - - 100


Vehicle type: Car



AV Advanced AV

Arterial C C AK

Urban street M C N


h2020-coexist.eu


Page 93 of 120


Vehicle type: Truck



AV Advanced AV

Arterial C C AK

Urban street M C N


Vehicle type: Bus



AV Advanced AV

Arterial C C AK

Urban street M C N


The relevant combinations of AV driving logics and functions are specified in Table 91. In the case of MK

the road types that would be considered would be arterial or urban street. Most roads external to the

centre and near the vehicle intercept locations on the edge of the centre would be classified as Arterial.

Within the vehicle intercept locations and within the centre roads are considered to be urban street in

behaviour.


h2020-coexist.eu


Page 94 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")







Parking Garage Pilot (Level 4) U


Traffic Jam Chauffeur (level 3) A U



Urban and Suburban Pilot (level 4) A U

Everywhere Pilot (level 5) A

Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning62


Possible



h2020-coexist.eu


Page 95 of 120

2 Traffic demand



The demand configuration describes the current demand. This demand configuration will only be used

for calibration and comparison.




Cars X

Trucks X

Buses X





h2020-coexist.eu


Page 96 of 120



Cars X

Trucks X

Buses X



locations.




Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


h2020-coexist.eu


Page 97 of 120



locations and their bottleneck capacities. This demand configuration will help in testing various detailed

designs for the vehicle intercept locations further.




Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


The demand configuration describes a prognosis in the case when vehicle and delivery access to the

city centre is restricted, vehicle intercept locations constructed, and services from vehicle intercept

locations into the centre provided. This will be testing the network effects with the introduction of the

aforementioned measures.



h2020-coexist.eu


Page 98 of 120



Cars X

Trucks X

Buses X




Cars X

Trucks X

Buses X


The demand configuration describes a prognosis in the case when vehicle and delivery access to the

city centre is restricted, vehicle intercept locations constructed, and services from vehicle intercept

locations into the centre provided. This will be testing the effect of the final measures any further

changes deemed necessary restriction access will be implemented.




Cars X

Trucks X

Buses X



h2020-coexist.eu


Page 99 of 120



Cars X

Trucks X

Buses X



for each stage of coexistence. The case with and without restriction of car traffic to the centre are studied



Demand level Demand 1 Demand 2 Demand 3 Demand 4 Demand 5

Introductory X X X X

Established X X X X

Prevalent X X X X






Sensitivity analysis would be undertaken as part of the use case to investigate the distribution of people

in private cars. Also during the designing of vehicle intercept locations sensitivity analysis will be

undertaken to asses boarding/alighting time of freight and its variance with different boarding/alighting

strategies such as boarding sides and gaps between vehicles.


h2020-coexist.eu


Page 100 of 120





Introductory X X

Established X X

Prevalent X X


h2020-coexist.eu


Page 101 of 120

Appendix I Use case 7: Impacts of CAV on travel time and mode choice on a network level


Version: 1.3

Date: 2018-04-04

Author: Markus Friedrich and Jörg Sonnleitner


for use case 7

Impacts of CAV on travel time and

mode choice on a network level





h2020-coexist.eu


Page 103 of 120

Table of contents

1 Level of automation ................................................................................. 104

1.1 Relevant combinations of CAV driving logics and functions ............................................ 105

2 Traffic demand ......................................................................................... 108

2.1 Traffic demand configurations ......................................................................................... 108

2.1.1 Demand configuration 1........................................................................................................ 108

2.2 Relevant traffic demand configuration per stages of coexistence .................................... 108

3 Traveller adaptation ................................................................................. 109


h2020-coexist.eu


Page 104 of 120


Table 15 presents indicative penetration rates of AVs for cars and trucks, which are the two vehicle types

included in the use case. The two main uncertain factors in this use case are the penetration rate of

various CAVs and the value of time for driving an AV. Cars and trucks are assumed to have the same

AV penetration rate.




Introductory 10-40%

Established 30-70%

Prevalent 60-90%

Table 17 presents the composition of AVs in terms of the AV classes: Basic, Intermediate, Advanced

(specified in Table 1 - Table 4). It would be preferable to vary the mix of AV classes but it will be difficult

for the macroscopic use cases. The mix is assumed to be fix in order to limit the number of microscopic

traffic simulation investigations required to derive passenger car units for the road environment

categories (motorways, arterial and urban street).





AV Advanced

AV

Introductory 80 20


Prevalent 20 80

Table 1 and Table 107 present a suggestion of according to which AV driving logic (rail-safe, cautions,

normal, all-knowing) that the different AV classes (basic, intermediate, advanced) are estimated to drive

according to.


h2020-coexist.eu


Page 105 of 120


Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C


Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C N

Shared space M M C


The relevant combinations of AV driving logics and CAV-functions for three typical road environments:

motorways (M), arterials (A) and urban streets (U) are specified in Table 4. Functions relevant for shared

space are not presented since the macroscopic model does not include shared space road

environments.

It is expected that a vehicle of the first generation of AV (Basic AV) will be able to handle situations on

motorways and on arterials on its own with high caution as well as the option to fall back to the manual

driver. For intermediate AV this capability will be extended to urban road environments. Additionally, the

included highway pilot doesn’t need a driver as a fall back on motorways anymore. Finally, the advanced


h2020-coexist.eu


Page 106 of 120

AV is capable to handle driving in all three road environments on its own, yet it is still not able to manage

all situations under all circumstances, which is why it is not a driverless vehicle.


h2020-coexist.eu


Page 107 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")






Park Assistance (level 2)

Parking Garage Pilot (level 4) A / U



Highway Chauffeur (level 3) M

Highway Pilot (level 4) M A / M

Urban and Suburban Pilot (level 4) U / A U / A


Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning70


Possible



h2020-coexist.eu


Page 108 of 120

2 Traffic demand



The demand configuration 1 which will be used within use case 7 complies with the forecasted travel

demand for the year 2025. The forecast is based on the travel behaviour of 2010. For this base year, the

travel demand model was calibrated and validated with observed travel behaviour. Furthermore, the

forecast includes all structural measures and changes regarding the population that can be expected by

2025.

Time period: static demand of a working day over 24 hours



for each stage of coexistence. Demand configuration 1 will be used for all three stages.


Demand level Demand 1

Introductory X

Established X

Prevalent X




h2020-coexist.eu


Page 109 of 120


The basic assumption is that traveller’s preferences are the same as today. However, the introduction of

CAVs could imply changes in traveller’s perception for example in terms of how they value their time.

Reasons for a change in the value of time are a higher comfort and the possibility to use the in-vehicle

time more efficiently, as long as the car takes over the control, which depends on the AV class and the

road environment.

The introduction of CAV-certified road sections or even a “geofenced” CAV-ready area will also lead to

an adaption of the travellers in terms of route choice, but this seems to be rather a measure than an

uncertain factor.

Table 31 describes the potential changes in traveller’s preferences that will be considered for this use

case.

Table 110 Traveller preference adaptation due to the introduction of CAVs for each stage of coexistence.


Value of time

Introductory X

Established X

Prevalent X

Another basic assumption is that non CAVs (vehicles, pedestrians, bicyclists, etc.) interact with CAVs in

the same way as they interact with manually driven vehicles today.

Free flow travel times will not change, as the Stuttgart Region travel demand model assumes that all

vehicles obey the speed limits.


h2020-coexist.eu


Page 110 of 120

Appendix J Use case 8: Impact of driverless car- and ridesharing services


Version: 1.4

Date: 2018-04-18

Authors: Markus Friedrich and Jörg Sonnleitner


for use case 8

Impacts of driverless car- and

ridesharing services





h2020-coexist.eu


Page 112 of 120

Table of contents

1 Level of automation ................................................................................. 113

1.1 Relevant combinations of CAV driving logics and functions ............................................ 116

2 Traffic demand ......................................................................................... 118

2.1 Traffic demand configurations ......................................................................................... 118

2.1.1 Demand configuration 1........................................................................................................ 118

2.2 Relevant traffic demand configuration per stages of coexistence .................................... 118

3 Traveller adaptation ................................................................................. 119


h2020-coexist.eu


Page 113 of 120


Use case 8 examines the impact of driverless car- and ridesharing services. Such services can operate

as independent mode offering direct connections or can be integrated into public transport as feeder

system (last mile service). Use case 8 makes assumptions concerning the future supply of conventional

public transport (bus, rail). The following table shows possible scenarios specifying combinations of

available modes.

Table 111 Combinations of available modes in use case 8

Subscenario

Public transport Car transport

Bus Rail RS+ NS CS RS-

S 8.0:

Public transport as today

No sharing

x x - x - -

S 8.1:

Public transport as today

CS & RS- as new competing modes

x x - x x x

S 8.2:

Public transport as today with RS+

CS as new competing mode

x x x x x -

S 8.3:

Rail transport as today

BRT supplemented by RS+


x (only

BRT) x x x x -

S 8.4:


BRT supplemented by RS+

CS & RS- as new competing mode

x (only

BRT) x x x x x

S 8.5:


No bus, only RS+


- x x x x -

S 8.6:

Public transport only with RS+


- - x x x -

Bus Buses operate according to a timetable with the same speed as today. No specific assumptions

on automation. Automation may reduce operating cost, if drivers are redundant.


h2020-coexist.eu


Page 114 of 120

Rail Light rail and heavy rail operate according to a timetable with the same speed as today. No

specific assumptions on automation. Automation may reduce operating cost, if drivers are

redundant.

RS+ Ridesharing integrated in public transport. This can come in the form of a last mile service or as

direct service in areas with low demand (bus on demand). The system may operate with one

vehicle size (e.g. 6 seats) or with a varying vehicle size. Passengers must transfer to bus/rail, if

this provides a reasonable service. Travel time depends on road saturation.

NS Private Car, no sharing, travel time depends on road saturation.

CS Carsharing, travel time depends on road saturation. This supply is equivalent to a personal on

demand (pod) service.

RS- Ridesharing not integrated in public transport. Travellers always travel without transfers. The

system may operate with one vehicle size (e.g. 6 seats) or with a varying vehicle size. Travel time

depends on road saturation.

BRT Bus rapid transit (major bus lines with high demand)

Table 15 presents indicative penetration rates of AVs for cars (including ridesharing vehicles) and trucks,

which are the two vehicle types included in the use case. Cars and trucks are assumed to have the

same AV penetration rate. The main uncertain factors in this use case are the willingness of the people

to share cars or rides as well as the fare model / prices for sharing services.

Table 112 Penetration rates of AV of vehicle types car and truck for each stage of coexistence.



Introductory -

Established -

Prevalent 100%

Table 17 presents the composition of AVs in terms of the AV classes: Basic, Intermediate, Advanced

(specified in Table 1 - Table 115). The assumption to have 100% driverless vehicles within this use case

implies to have a penetration rate of 100% as well as 100% of the most highly developed AV. The

complexity of the use case 8 derives not from various mixes in penetration rate and AV fleet

compositions, but from the variety of possible mobility services and additional measures.


h2020-coexist.eu


Page 115 of 120

Table 113 Indicative ratios of AV of vehicle types car and truck for each AV class for each stage of coexistence.




AV Advanced

AV

Introductory

Established

Prevalent 100


Vehicle type: Car



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C C

Shared space M M C


Vehicle type: Truck



AV Advanced AV

Motorway C N AK

Arterial C C AK

Urban street M C C

Shared space M M C


h2020-coexist.eu


Page 116 of 120



motorways (M), arterials (A) and urban streets (U) are specified in Table 4. It is expected that all vehicles

can operate driverless in all situations.


h2020-coexist.eu


Page 117 of 120



RS C N AK

VR

A f

un

cti

on

s

(acc

ord

ing

to

th

e p

ap

er

"D

ep

loym

en

t p

ath

s

for

Ve

hic

le a

nd

Ro

ad

Au

tom

ati

on

")






Park Assistance (level 2)

Parking Garage Pilot (level 4)






Everywhere Pilot (level 5) U M / A

Co

mm

un

icati

on

fun

cti

on





Co

op

era

tio

n

fun

cti

on




Platooning78


Possible



h2020-coexist.eu


Page 118 of 120

2 Traffic demand



The demand configuration 1, which will be used within use case 8, complies with the forecasted travel

demand for the study area for the year 2025. The forecast was estimated with the travel demand model

of the study area based on the travel behaviour of 2010. For this base year, the travel demand model

was calibrated and validated with observed travel behaviour. The forecast includes all structural

measures and changes regarding the population that can be expected by 2025, but does not yet

consider CAV. The demand configuration does there not include any induced demand which may arise

from CAV or need vehicle sharing systems using CAV. Impacts of CAV on demand are described in

chapter 3.

For use case 8, the daily travel demand will be computed in 96 time slices of 15 minutes. This time-

dependent demand serves as input for vehicle scheduling and possible relocation of vehicles within

sharing services.

Time period: dynamic demand of a working day over 96x15min



for each stage of coexistence. The demand configurations will only be used for the prevalent stage, in

which it is assumed to have 100% Advanced AV being able to operate driverless.




h2020-coexist.eu


Page 119 of 120


Demand level Demand 1

Introductory

Established

Prevalent X


The basic assumption is that traveller’s preferences remain the same as today. However, the

introduction of CAV can bring changes two types of changes in the behaviour of travellers:

Perception of travel time: Car-drivers can use their in-vehicle time more efficiently.

New choices for travellers without car availability or driving licence, who can now use a private

car or a shared car similar to people with car availability.

The introduction of private or public vehicle sharing services requires the introduction of a fare model to

compute the prices of shared modes. The resulting prices influences the behaviour of the travellers and

will be analysed in a sensitivity analysis with varying fares.

Table 31 describes the changes considered in use case 8, which will lead to an adaption of travel

behaviour The changes include traveller’s preferences of the mode car, new modes and the prices of the

new modes.

Table 118 Traveller adaptation due to the introduction of CAV for each stage of coexistence.


Car Availability New modes Prices

Willingness to replace private car with a shared car

Introductory

Established

Prevalent

mode “car driver”

also available to

person groups

without car

availability

carsharing

direct ridesharing

ridesharing

integrated in

public transport

price sensitivity

remains fixed in

the utility function

prices for trips

with new shared

modes influence

mode choice

X


h2020-coexist.eu


Page 120 of 120

Another basic assumption is that non CAVs (vehicles, pedestrians, bicyclists, etc.) interact with CAVs in

the same way as they interact with manually driven vehicles today.

important note - coexist · 2019-10-31 · important note this coexist deliverable is part of a...

Documents