Behavioural aspects of automatic vehicle guidance : the relationship between headway and driver comfort

Automation of road traffic may have a large potential for improving the performance of the traffic system. For the not too distant future, systems that support or automate parts of the driving task will appear on the market. On a longer term fully automated driving on parts of the road network may become possible. The acceptance of automated driving may play an important role in the feasibility of Automated Vehicle Guidance (AVG). As short headways may result in the most dramatic increase in road capacity, the Ministry of Transport, Public Works and Water Management has commissioned the TNO Human Factors Research Institute to carry out a study into the acceptability of short headways in an automated traffic system compared to manual driving. The experiment consisted of two parts with complementary methods to assess the relationship between headways and acceptability. In one part subjects gave a subjective comfort rating about the condition they were driving in, while in the other part subjects were allowed to adjust the headway to a comfortable level. Furthermore subjects drove the simulator manually before and after each AVG sessions. In the manual sessions, subjects rated the comfort level in a number of traffic conditions.
Based on the results of the experiment, recommendations were derived for the design headway of a comfortable AVG system. In order to equal the comfort level in dense traffic as experienced daily on the motorway network in rush hours, the AVG headway should be no less than 0.86 s. If it is accepted that the comfort level that occurs in incident situations would already suffice, the AVG headway could be 0.29 s.
An increase of driver comfort at very short headways was not found in this experiment (a u-shape of the comfort versus headway curve).
The preferred headways correspond to the values observed in normal traffic. Starting at a very short headway the average comfortable headway was 0.70 s, while when starting at a long headway the preferred headway was 1.46 s. On average a headway of 1.1 s was adopted.
An increase of the free driving speed after the experimental AVG sessions was found. Whether this was merely a result of driving (a simulator) for some time or whether this effect was caused by prolonged driving in AVG should be investigated in a separate experiment, including a control group that drives manually during a comparable period.