Karel Brookhuis (University of Groningen, Delft University of Technology) gave a presentation on human factors evaluation of automation. He started his presentation with a figure that states that 85% of the crashes is directly attributable to the driver (error, impairment). Benefits from Advanced Driver Assistance Systems (ADAS) are to be expected. Consider for example a decrease in accidents, fuel usage and pollution, an increase in road capacity and in total a decrease in societal cost. However, the human factor, i.e. the driver, may be a problem. Automation changes the driving task from driving towards supervising the car.

Three operation modes or functionalities can be distinguished:


Advisory – information provision, e.g. route guidance, ISA;


Semi-automatic – active support, taking over part of the control, e.g. ACC, ISA;


Automatic – taking over control completely, e.g. Path, Phileas.

In the presentation three human factors evaluation studies are highlighted. These studies were performed with a driving simulator and concerned ACC, Path and Phileas.

Benefits of ACC are considered to be a decrease in human error (leading to less accidents), an increase in efficiency (by a decreasing headway) and a reduction in fuel use. However, potential problems can be liability in case of an accident, complacency and behavioural change. This last issue is investigated in a driving simulator study with 40 drivers that differed from each other with respect to speed preference (high/low) and focus (high/low). Results indicate a reduction in minimum time headway, especially for low speed drivers. Behavioural changes are mostly found for high speed drivers: more left lane driving, more overtaking and more weaving.

After this example of a human factors evaluation study on a semi-automatic system, Brookhuis continues his presentation with discussing whether driving should be made completely automatic. When the vehicle takes over control, no (active) driver is involved, resulting in no human errors and thus saving costs. Problems might be, however, acceptance, loss of skills and liability in case of an accident. Human factors were studied in a driving simulator experiment in which an emergency situation on an automated highway was simulated. It was investigated how drivers of an automated Path vehicle would react if a vehicle cuts in from the emergency lane. Half of the drivers showed no reaction at all, they were only supervising the car. Only very few drivers braked early in order to avoid a collision. People have certain expectations of a system (e.g. the automated vehicle will deal with the car cutting in), which not always correspond with its possibilities and limitations.

A third human factors evaluation study concerned automatic public transport. Phileas is an automated bus that can automatically follow a predetermined trajectory, using for example an electronic lane assistance and precision docking system. In 2004 the Phileas bus will be operative in Eindhoven, the Netherlands. Relevant aspects of automatic bus driving are the different driving task (supervising instead of driving) and the need for trust, believe, acceptance. Driving behaviour and acceptance were studied in a driving simulator experiment with 25 bus drivers. Two incidents were simulated: (a) car blocking the road and (b) cyclist running red light. The drivers reacted well on the first incident by braking ample in time and returning to the half-automatic driving mode. With respect to the second incident, however, almost a quarter of the drivers did not react on time. But the second time they experienced this scenario, everybody reacted in time. Driver’s attention may decrease when (parts of) the driving task are automated; drivers are inclined to ‘sit back’. However, when drivers were explicitly told to supervise and interfere when necessary, the attention level for unexpected events increased. Training on unexpected events (preferably in a driving simulator) seems feasible. Attention should be given to adapted, specific and not everlasting driving licenses and periodic training.

After the presentation some questions were posed. It was asked for which systems drivers should need a special driving license. Such a driving license should be necessary for each system that has effects on the driving behaviour, i.e. that induces behavioural adaptation. With respect to suggestions to keep drivers active when using ADAS, it was mentioned to train people to anticipate and prepare for unexpected situations. How long a training with ADAS should last is not known at this moment. In the Phileas study two and a half hours were used for training and this seemed to be long enough. Next year a study will be undertaken to investigate the optimum duration of training. However, a problem is (still) finding suitable criteria for (un)safety.

Click here for the slides of the presentation.

More information on the topic, written by Karel Brookhuis, can be found here:


Behavioural responses to Intelligent Transport Systems


De interactie tussen techniek, beleid en gedrag in verkeer en vervoer [in Dutch]