On the interaction between driver assistance systems and drivers in situations of system failure

Licentiate thesis, 2011

The level of automation is increasing in road vehicles as driver assistance systems take control of both speed and steering. Benefits of this development are expected in the form of improved safety, driver comfort, traffic efficiency, etc. However, with increased technical complexity, software content, and electro-mechanical interfaces, there are also increased risks from failures. The objective of this thesis is first to investigate driver performance in situations where assistance systems fail, and second to propose strategies to improve these situations by optimizing driver controllability. In this work, an adaptive cruise control system, a system that automates speed keeping of the vehicle, was chosen as a representative example of a driver assistance system for analysis. One driver interview study and one experimental driving simulator study was conducted. Results indicate that drivers' use of assistance systems change after time of usage and that their interpretation of the functionality of the systems can be rudimentary even after several months of usage. Because an adequate and correct understanding of the systems is important to control failures, it is proposed that drivers are informed and educated about system capabilities and limitations. Furthermore, a moving-base driving simulator was developed and used in an experimental study to investigate drivers' performance when an adaptive cruise control system fails. Findings suggest that patterns can be identified for the strategies that drivers apply to control failures. It was also found that certain failures are more difficult for drivers to control than others.