Summary_JH

Jeroen Hogema (TNO Human Factors) gave a presentation on driving behaviour effects of the Chauffeur Assistant. As part of the EU project CHAUFFEUR II a combination of a Lane Keeping System (LKS) and Adaptive Cruise Control (ACC), named Chauffeur Assistant, is developed for trucks. Under the authority of the Ministry of Transport the project Dutch Evaluation of the Chauffeur Assistant (DECA) is performed.

The main questions within this project concern the impacts of the Chauffeur Assistant (CA) on:

-

the individual level, in terms of driving behaviour, workload and acceptance

-

the traffic flow level, in terms of traffic performance and safety indicators.

By means of a driving simulator experiment with the TNO truck simulator the impacts on the individual level have been determined. The driving behaviour results are incorporated in the microscopic simulation model MIXIC in order to determine the impacts of the Chauffeur Assistant on the traffic flow level. In the presentation the driving simulator experiment and the consequences of its results for MIXIC were highlighted.

Eighteen professional truck drivers experienced the CA on a simulated 3-lane motorway with 3.5 m wide lanes. Prior to the experiment each driver selected a preferred ACC headway setting. The subjects drove with and without the CA in situations with low and high traffic volume. The CA could be turned on by using a switch and turned off by using the switch or brake pedal. Two scenarios were presented: (a) car-following (overtaking possible) and (b) braking lead car.

ACC headway settings of 1.3 s and 1.6 s were most preferred by the truck drivers. When a lead car is braking fewer lane changes are undertaken when the CA is used. In this scenario also a lower deceleration level is achieved when the CA is used. The acceptance of the CA is rather high: on a scale from –2 up to 2, usefulness is valued at +0.93 and satisfaction at +1.10. However, the rating on ‘raising alertness – sleep inducing’ needs attention, because most drivers thought the CA to be sleep inducing. The results of using the CA can be summarized as follows:

-

Reduced SD of lateral position;

-

Higher Time-to-Line-Crossings;

-

Less short time headways;

-

Reduced mental effort (measured by the RSME);

-

Fewer lane changes (when a lead car is braking);

-

Less critical behaviour in terms of maximum deceleration and minimum Time-To-Collision (when a lead car is braking);

-

The individual level, in terms of driving behaviour, workload and acceptance;

-

The traffic flow level, in terms of traffic performance and safety indicators

No effects were found on the mean and SD of speed, lane use (% right lane), mean lateral position and mean time headway.

The results of the driving simulator experiment are incorporated into MIXIC in order to assess the impacts of the CA on the traffic flow level. The driver model of MIXIC consists of three elements: longitudinal (car following, free driving), lateral (lane change model) and the driver support system (CA: settings and transitions). The driving behaviour results have particularly consequences for the simulation of the CA in MIXIC:

-

CA reference speed = driver’s intended speed;

-

CA reference headway: 50% 1.3 s; 50% 1.6 s;

-

CA off when (1) CA is braking hard AND driver would brake harder, or (2) driver starts lane change manoeuvre;

-

CA on when “possible”.

Some small effects on lane change behaviour were found when the CA is used. However, the nature of these effects is unknown (e.g. reduced alertness?). Therefore, no changes are made in the lane change model of MIXIC.

It was concluded that the effects of the CA on behaviour, workload and acceptance are in line with ACC or LKS research. The precise contribution of ACC and LKS is unknown, because these features were only studied together. Based on the results of the driving simulator experiment some minor modifications to MIXIC are made to assess the impacts of the CA on traffic performance and safety.

After the presentation several questions were posed. With respect to the design of the experiment it was asked why the CA (in particular the ACC) is off when the driver is changing lanes. This is due to the specifications for the system that were made by DaimlerChrysler. The CA is either completely on or completely off. It was mentioned that time and money restricted the choice of scenarios in the experiment. In this experiment the impacts of the CA on the individual level in normal situations instead of more critical and unintended scenarios were investigated. The truck drivers were allowed to turn off the CA during the experiment, however, they were told to do this as little as possible. Therefore it is not known to what extent drivers would really like to have such driver support.

Another question concerned possible compensating behaviour of drivers when using the CA. It was found that most drivers accepted the reduction in workload, thus relaxing and supervising the system and spending more time on braking (before the system does). This probably caused the less critical behaviour with the CA when a lead car was suddenly braking. It was not investigated in this short-term experiment whether the drivers were willing to change their time headway after they had used the CA. However, in the long-term it is possible that drivers would like to change the time headway, because they get a better mental model of the system. Already an indication for the formation of mental models was found in the experiment, in terms of knowing the limitations of the system and compensating for them. However, not everybody in the audience agreed with that: drivers are not able to form a mental model after one short driving simulator experiment. The effects of a platoon equipped with the CA are part of the MIXIC study. It has not been investigated in the driving simulator experiment, because only one vehicle was equipped with the CA. In the MIXIC study 10% of the vehicles will have a CA system.

Click here for the slides of the presentation.