Effectiveness Evaluation of Simulation-Based Training for Potential Drivers

 

Tova Rosenbloom & Ehud Eldror

Bar Ilan UniversityIsrael

The research was financed by the National Authority of Road Safety

Tova Rosenbloomtovarosenbloom@gmail.com

Tel: 972524449998

Simulators as a part of Driving Instruction

• Driver’s age is one of the variables that is 

strongly related to the higher rate of road 

    crashes (Turetschek, 2006). 

• The technical skills of novice drivers are lower 

than those of older drivers (Clarke, Ward, 

Bartle, & Truman, 2005; Ulleberg & Rundmo, 

2002)

• Thus, early educational and training programs for the reduction of risky driving are important even at the onset of driving lessons (McKnight & Peck, 2002;Morrisey, Grabowsky, Dee, & Campbell, 2006).

• There are many intervention programs to promote safe driving of novice drivers.

• Some of these programs are based on simulated situations (Williams, 2006).

• The present study is aimed at investigating the benefits of a simulator-based training program for the safe driving of novice drivers as part of their mandatory driving lessons.

Advantages of Simulator-Based Driving Learning

• Simulators enable computed variations of dangerous driving situations in various road conditions can contribute to the practice of novice drivers.

• Simulators enable large range of road and traffic conditions that can be represented (light/darkness, different weather conditions, various infrastructure conditions and interactions with other road users, etc).

• The simulations can also be adjusted to the needs of various groups of students (Deery & Fildes, 1999; Deery, Kowadlo, Westphal-Wedding & Fildes, 1998).

• According to this approach, it can be hypothesized that theoretical driving studies combined with practical training on simulators will elevate the safety level of novice drivers (Slick, Cady, & Tran, 2005). 

Disadvantages of Simulator-Based Driving Learning

• The generalization of the outputs of simulator use to everyday driving is problematic. Christie (2000), and Mayhew and Simpson (1995) assert that training in the real environment is preferable than with a simulator

• Godley, Triggs, and Fildes (2002) report differences between driving speed in a simulator and in real driving, explained by the higher sensation levels that real driving provides for the driver.

• Feelings of drowsiness in simulator are higher than in real driving (Philip et al., 2003)

• People become fatigue after shorter time of driving in simulator than in real driving possibly due to a lack of stimuli (George, 2003). 

• Previous research indicates that drivers who were trained with a simulator were involved in a lower rate of road crashes than those drivers in a control group (Allen, Park, Cook & Fiorentino, 2007).

• Thus till now, the findings of most of the studies suffer from methodological limitations or were based on simulated driving parameters  as dependent variable and not on real driving parameters.

• The present study attempts to support direct evidence of the advantage that simulator-based training has compared to non-simulator-based training in reference to safe driving.

The present study compared novice drivers of two groups (trained with simulator/trained without simulator) in driving parameters that included self-reported intentions of safe driving, theoretical driving knowledge and real driving.

The Present Study Compared:

• Self report of behavioral intention of safe driving

• Theoretical knowledge of driving• Attitude towards safe driving• Objective and subjective assessments of real

drivng

Research Design

The Studyn=280

Part BParameters of Real

Drivingn=40

Part AParameters Gathered by

Questionnairesn=280

Expert AssessmentIVDR

KnowledgeSelf Report on Behavior

The basic assumption:

Safe Driving id typified by:•Safer attitudes•More knowledge •Lower and more stable driving parameters•Lower variance of vehicle parameters•Proactive (versus reactive) style of driving

Part A: Questionnaire Study Participants:•280 novice drivers living in the city of Ashdod •140 that have been trained with simulators•140 that have been trained without simulators

Part AInstruments:

Questionnaire - Based on TPB (Ajzen, 1991)

Attitude toward safe

driving

Subjective Norm

Perceived Control

BehavioralIntention Behavior

Part A

The Questionnaire related to:•Driving under influence of alcohol•Speeding•Driving in fatigue

•The higher the score the higher the safe measurement

Part A

Examples for questions:• I intend to drive only at the legal speed

(Intention)• My parents and friends expect me not to

speed (Subjective Norm)• The decision whether to drive after alcohol or

not is absolutely mine (Perceived Control)• Driving by the rules is very relaxing (Attitude)

Part A

Knowledge Questions:

Twenty-two questions taken from theory exams such as:

In which case it is allowed to make u-turn?

1. Anywhere

2. When there is a lane for turning to the left and there is no sign that forbid it

3. When there is no sign that forbid it

4. All the answers are wrong

Part A - Results• Unsafe Driving Intention

With Simulator > Without Simulator

.15, p < .005-β = = .02, R2 = .47 ΔR2

• Speeding Intention

With Simulator > Without Simulator• .16, p < .005-β = = .02, R2 = .45 ΔR2

• Knowledge

No Difference

Part A - Explanations

• Unsafe Driving Intention • Speeding IntentionWith Simulator > Without Simulator

• overconfident approach towards speed driving in drivers who received simulator training, not unlike that which was found for newly licensed drivers who received skid-training (Glad, 1988; Gregersen, 1996-spelled differently in bibliography; Jones, 1993).

It may be that drivers who received simulator training consider their high speed driver skills to be better. Interestingly, while both personal attitudes and perceived control were associated with safer driving intentions for driving while intoxicated, subjective norms were not.

The lack of a subjective norms effect may indicate an awareness of the effect of peer pressure on driving while intoxicated or a form of social desirability.

Part A

• Theoretical traffic knowledge was similar in both groups regardless of simulator training, suggesting that the theoretical knowledge quality was independent of the nature of the actual driving training, be it limited to on-road training or combined on-road and simulator training.

Part B – Real Driving Evaluation

Participants:•Of the 140 drivers who received simulator-based learning 20 drivers were invited to this phase.•In parallel, of the 140 drivers who did not receive simulator-based learning 20 drivers were also invited to participate in this phase.

Part B

Instruments and Apparatus• Driving expert evaluation. A professional driving

instructor rated the driving performance of each driver participating in the real driving evaluation part.

• The drivers were rated on a list of pre-determined parameters pertaining to vehicular control, traffic merging and road use.

• Scores ranged between 0, representing unsatisfactory performance and 4, representing extremely satisfactory performance.

Part B

• Objective measures of real driving• Mobileye AWS-400 System. This system

includes a camera unit, a control and feedback unit and two speakers that provide real time video-processed monitoring of the vehicle’s relative lane position.

The system measures headway and driving lane deviations via the camera unit as well as monitoring speed and braking data via the vehicle sensors.

The sampling rate for the camera data is 10 samples per second. The headway range is up to 70 meters on a 40° x 30° field of view, with a two seconds time to collision threshold.

Part B

• TrackTeck DPL (Dual Purpose Locator) V3 system. This system is comprised of a GPS positioning system and a G-sensor monitoring the vehicle acceleration. Data sampling rate is 10 samples per second.

Part B

• Traffilog System. This system provides real time monitoring of both the vehicle sensors and a g sensor, displaying driving characteristics pertaining to acceleration, turning and braking events according to preset criteria, informing of the severity of each event, which has been defined by the level of deviation of each event of the total events.

Acceleration events include five levels of severity, of which the top two are defined as most unsafe events. Turning events include eight levels of severity, of which the top three are defined as safety events.

Braking events include ten levels of severity of which the top three are defined as safety events. Threshold and event definition algorithms were predetermined by the manufacturer.

Part B

In sum, the data recorded from the three systems included:• mean driving speed (KmH)•driving speed variance•mean driving distance•amount and duration of brake pedal presses•time to collision•headway events count•the count and the level of deviation of acceleration, turning, and braking events.

Part B

• Procedure• Driving evaluation• Each of the 40 drivers participating in the driving

evaluation drove the IVDR-equipped Toyota Corolla accompanied by a driving expert through a pre-determined course within and beyond the limits of the city of Ashdod.

• The driving experts provided the drivers with route instructions. The driving sessions were given on Wednesdays and Thursdays, from11:30 to 17:30 during the months of November and December 2011, in good weather.

Part B

The driving expert was unaware of the purpose of the study and of the drivers’ original training (i.e. with or without simulator).

Part B

The IVDRs monitored the sessions with their feedback apparatuses turned off and the drivers received no feedback given on real time of their driving.

Part B

Driving expert evaluation•T-tests were conducted for comparing the driving expert’s evaluation of the actual driving performance of drivers who received and did not receive simulator training. •Comparisons failed to yield any significant difference between the drivers’ groups, both for the overall evaluation score and for the specific evaluation categories.

Part B

Objective measures• Comparisons of the mean scores for the

various monitoring systems data were performed using t-tests.

• Data included severity levels for discrete accelerating, turning and braking events as well as the mean braking pedal press duration, and the mean and variance of headway distance.

Part B

• Additionally, t-test comparisons of the driving distances, average driving speed and its variance were performed in order to confirm the absence of qualitative differences between drivers’ sessions.

• Comparisons failed to yield significant differences between the drivers’ groups in any of the monitored measures.

Part B

• Poisson regressions comparisons of the number of events registered on the various monitoring systems were performed, which included the number of turning, accelerating and braking events, headway events, and brake pedal presses count.

Part B

Brake pedal press count was significantly higher for drivers who received simulator training compared to drivers who did not receive simulator training.

β Wald's χ2 SE OR LR χ2

-0.15 839.88* 0.01 0.87 *839.95

Additionally, the headway events count was significantly lower for drivers who received simulator training compared to drivers who did not receive simulator training.

β Wald's χ 2 SE OR LR χ 2

0.16 260.00* 0.01 1.18 261.24*

No other significant differences were found.

Part B

Thus, the driving of those participants whose training was simulator-based was most reactive than those who were not trained with simulator.

Conclusion

The present research did not indicate any crucial difference that could demonstrate on advantage or disadvantage of combining simulators in driving instruction.

Thank you!