towards an objective framework to assess the … · towards an objective framework to assess the...

25
Towards an objective framework to assess the impact of driverless vehicles Dr. Sven Maerivoet 1 Bits ‘n Chips | 1 October 2015

Upload: lamquynh

Post on 04-Aug-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

Towards an objective framework to assess the impact of driverless vehicles

Dr. Sven Maerivoet

1 Bits ‘n Chips | 1 October 2015

Transport & Mobility Leuven (TML)

● Founded in 2002 as a spin-off company (NV):

– KU Leuven TNO research institute (The Netherlands)

– Multidisciplinary team (18 people)

– Independent and open policy (cf. www.tmleuven.be)

– Clients: cities and regional governments, Flanders, Belgium, Europe

– Relations: ITS Belgium (Board), VIM (advisory council), EC 7KP, …

● Quantitative and policy supporting: – Traffic flow theory (incl. ITS measures, congestion estimation)

– Transport economics (incl. impact assessments, SCBAs, …)

– Private road, rail, public transport, inland waterways, air, …

– Environment, public health, traffic safety (incl. legislation, infrastructure, vehicle technology, …)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 2

Introducing TML

Company|Press|Congestion

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

TML is regularly featured in the press

● Het Nieuwsblad, De Morgen, De Standaard, Het Laatste Nieuws, Knack, Jobat, Le Soir, La Meuse, De Lloyd, Verkeersspecialist, Vacature Magazine, NM, Ademloos, Het Belang van Limburg, Mobimix, De Zondagskrant, De Streekkrant, Verkeersnet.nl, …

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 3

Introducing TML

Company|Press|Congestion

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

Background: some numbers on congestion

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 4

Introducing TML

Company|Press|Congestion

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

Autonomous Vehicles

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 5

The first signs of autonomy

● Radio controlled in 1926!

● 1980 – 2003:

– Strongly dependent on infrastructure

● From 2004 on:

– DARPA Grand Challenge

– 240 km in the Mojave desert

– Heavily equipped vehicles

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 6

(The Milwaukee Sentinel, 1926)

Introducing TML

Autonomous vehicles

Start|Legislation|Levels|V2X

Expected impacts

Accident & Behavioural Framework

The long run to autonomous vehicles

● Advanced Driver Assistance Systems (ADAS)

1. Safe speeds and following distances

2. Lane guiding

3. Detection of obstacles and collision avoidance

4. Satefy of intersections and complex situations

Tools!

● Autonomous driving

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 7

(A Quick Scan of Quantified Effects of ADAS – ADASE II Extension, 2004)

Introducing TML

Autonomous vehicles

Start|Legislation|Levels|V2X

Expected impacts

Accident & Behavioural Framework

Evolution in legislation

● Locomotives ‘Red Flag’ Acts (1865)

– Max 3 km/h in cities

– Required: driver + stoker + flag

– Background: protection of rail and horse carriage industries

● Vienna Convention wrt. road traffic (1968)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 8

Introducing TML

Autonomous vehicles

Start|Legislation|Levels|V2X

Expected impacts

Accident & Behavioural Framework

Different levels of autonomy

● Level 0 (no autonomy):

– Driver always takes actions

● Level 1 (driver task support):

– Steering or deceleration/acceleration are controlled

● Level 2 (partial autonomy):

– Steering and deceleration/acceleration are controlled

● Level 3 (conditional autonomy):

– Surroundings + intervention of the driver is still required

● Level 4 (high autonomy):

– Partial control (even if driver does not interfere)

● Level 5 (full autonomy):

– Full control of all driving tasks

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 9

!!! National Highway Transportation Safety Agency (NHTSA , 2013) Society of Automotive Engineers (SAE, 2013)

(SAE, 2013)

Introducing TML

Autonomous vehicles

Start|Legislation|Levels|V2X

Expected impacts

Accident & Behavioural Framework

(Fleet renewal every 16 year)

Autonomy with integration of V2V and V2I (V2X)

● VANETs (mesh grids) + cooperative driving

● Communication with (intelligent) intersections

● By the way: Google Car contained a very detailed map

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 10

Introducing TML

Autonomous vehicles

Start|Legislation|Levels|V2X

Expected impacts

Accident & Behavioural Framework

Expected Impacts

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 11

Impact op stability of traffic streams: positive

● Increased stability because:

– Faster reactions of AVs

– Congestion seeds become scarce

– In principle: higher density of traffic stream

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 12

(Treiterer and Myers, 1969) (Maerivoet, 2005)

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Impact on throughput and jam formation: uncertain

● Vehicles can driver closer to each other – Platooning (although commercial OEMs go for longer distances)

– Better interaction and synchronisation of key manoeuvers during overtaking, at on- and off-ramps, …

– Shift towards higher throughput

– Strongly dependent on the mix of AVs (non-|semi-)AVs

● Cost of congestion: – Internal cost diminishes: our time is better spent

– External cost rises: if #vkm increases because driving becomes cheaper [rebound effect]

● A thought: are AVs as comfortable as, e.g. trains? – Requires lower accelerations

– Can lead to a capacity reduction (on the order of 20% – 50%)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 13

(The Tension Between Autonomous Cars’ Impacts on Junction [Intersection] Capacity and Occupant Comfort, 2014)

(Autonomous Vehicle - A Guide for Policymakers, 2014)

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Impact op traffic safety: strongly positive

● Mostly already a consequence of existing ADAS

● Inspired on Heinrich’s safety pyramid:

– Heavy + licht: vehicle construction + autonomy

– Almost: traffic education + follow-up/enforcement

● Attention with mix AVs (non-|semi-)AVs

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 14

Zwaar

Licht

Bijna

Zero-fatality is virtually unattainable

(Road Safety with Self-Driving Vehicles: General Limitations and Road Sharing with Conventional Vehicles, 2015)

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

H.W. Heinrich (1939), Industrial Accident Prevention

Ethical aspects related to traffic safety

● Suppose: an AV calculates that a crash is inevitable

● What choices are being made? – Discuss the moral dilemma

– Where lies the responsability: owner/passengers, insurance companies, constructors, road managers, …

● Luckily AVs have much higher reaction times than humans

● Standardisation wrt. safety & control: – Automotive Safety Integrity Level (ASIL): levels A – D

– Probably inclining towards ASIL-D

● Phenomenon of ‘hacked cars’ (cut breaks, shutdown, unlock, systems take over, sensor phantoms, …)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 15

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Impact on passengers: uncertain

● Psychological impact:

– When drivers becomes passengers…

– … they might not want the vehicle to drive as fast

● Research needed/ongoing wrt.:

– Confidence building with ex-drivers

– ‘Humanising’ AVs

– Cf. comfort level becoming a passenger

– The lost joy of driving

– ‘Offset’ behaviour: because of more autonomy, you feel more safe and might be inclined to take more risk

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 16

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Impact op traffic management: positive

● Security versus privacy, remote police control, …

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 17

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Impact on other aspects

● Impact on rural land use: – Living further away / prepared for longer travel due to more

quali-time

● Impact on cities: – Less parking problems (‘drop-off and continue’)

● Impact on energy and emissions: – Probably less (even with an increase in the #vkm)

– Green waves, more efficient fuel consumption, …

● Impact on costs: – Uncertain due to complex problem

– Actors: owner, parking-/road manager, jobs (taxis, PT, …), …

– Probably positive effect on society

● Impact on personal mobility: – An AV can be used by anyone (even by people with less to no

driving experience)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 18

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

The way forward!

● A government legally facilitates trials and validation tests

● Now often on private terrains: shuttle services (cf. airports), industrial terrains, agricultural sector, …

● Lommel test circuit public road

● Onset of trials with PT (De Lijn, Flanders’ MAKE, and VDL) and taxis

● Potential seen by Uber and Google

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 19

Introducing TML

Autonomous vehicles

Expected impacts

Stab.|Jams|Saf.|Pass.|Mgmt.|*

Accident & Behavioural Framework

Accident and Behavioural Framework

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 20

EMDAS (collective of Flemish companies/research institutes)

● Goal: develop a longitud. self-driving bus (v < 30 km/h)

● TML develops an objective framework for the assessment of safety impacts:

– Compare accident risk with and without AVs

– Change in accident risk ifo. the level of autonomy

– Understand the interactions between driver and vehicle

– Insight into unsafe behaviour that follows from this

● Need on a dedicated analysis of accidents

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 21

Introducing TML

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

EMDAS|Schemas|PreScan|X

Analyse common accident schemadata (CADAS / GIDAS)

● Derive accident typologies

● Validate with existing research (BIVV, VSV, IMOB, …)

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 22

Introducing TML

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

EMDAS|Schemas|PreScan|X

We get ‘equipped vehicles’

Simulation test

Provide feedback (sensor specs.)

Creating PreScan experiments

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 23

Introducing TML

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

EMDAS|Schemas|PreScan|X

Beyond longitudinal control: safety at intersections

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 24

Introducing TML

Autonomous vehicles

Expected impacts

Accident & Behavioural Framework

EMDAS|Schemas|PreScan|X

More information?

● Transport & Mobility Leuven:

– http://www.tmleuven.be/

[email protected]

– +32 16 31 77 33

01/10/2015 | Bits 'n Chips | Impacts of Driverless Vehicles 25