Nuisance vs False Collision Warnings During Driving; The Difference in Effect of False and Nuisance Alarms on the Driver when Driving with a Forward Collision Warning System

Abstract

As a diverse field of advanced driver assistance systems (ADAS) are developed and incorporated into the driving environment the need for analysis of their effects on the driver becomes increasingly important. As ADAS gain more influence on the driving task they change how we drive with a great potential for increasing the safety of the driver and those in his/her vicinity. There is however also a potential problem with the increasing amount of aids, if tuned incorrectly they may decrease safety instead; it is therefore of vital importance that the effects of aids and how they relate to potentially hazardous situations is well understood. This thesis strives to provide a contribution to the research field of ADAS in order to increase safety in the driving environment, specifically in the area of trust in frontal collision warning (FCW) systems under dif- ferent circumstances. Tests were carried out in a solid base driving simula- tor where reaction times were measured, followed by subjective measure- ment of workload by means of NASA-TLX questionnaire and subjective ratings of trust by means of a novel questionnaire based on the APT- framework[12]. The test was between groups by design where one group drove a sce- nario using a FCW system prone to giving nuisance alarms and the other group a similar scenario with a FCW system prone to false positive alarms. The FCW used was similar to that used in Volvo's cars. Both scenarios presented the test subject with 6 alarm situations, of which two where dif- ferent between the scenarios. Each scenario proceeded for 12-14 minutes depending on speed and reactions to events. No significant differences was found between the two groups in any area of data extracted. It is concluded that there is no difference in impact of nuisance and false positive alarms in either of the areas of performance, workload or trust as they were constructed here. There is evidence of the participants rating the type of alarms differently between the groups, supporting this conclusion. Reaction times lie well within what can be considered normal during real driving, indicating that the data found can be useful in further studies in the area of ADAS. The results of this study needs to be further tested but it carries some implications for the algorithms used in FCW systems in what types of alarms should be triggered. A proposed continuation of this study is in the relation between true positive alarms and nuisance alarms to further define the role of nuisance alarms, as well as a longitudinal study showing the long term effects of nuisance alarms.