Dealing with Uncertainty in Automotive Collision Avoidance
2004 (English)In: Proceedings of the 8th International Forum on Advanced Microsystems for Automotive Applications, 2004, 165-175 p.Conference paper (Refereed)
Modern automobiles incorporate more and more active driver assist system that uses sensors and microprocessors to control the vehicles dynamics. This paper discus a driver assist system that performs autonomous braking when the vehicle is close to colliding. Decision making in such systems is inherently uncertain, due to the sensors’ measurement uncertainty and the uncertainty of the driver’s future actions. Furthermore computational capacity is limited by the microprocessors used in automotive applications. In this paper considerations for dealing with the uncertainty of the estimated parameters in the decision making process is discussed. Risk metrics and a computationally efficient method for decision making is proposed. It will be shown that under certain conditions the risk for a to early intervention can be kept constant for different closing velocities using the proposed method. Furthermore tracking and modelling of driver, sensors and brake system is discussed. The models of driver actions and of the radar sensor measurement error are based on measurements from a Collision Avoidance system equipped Volvo V70, provided by Volvo Car Corporation.
Place, publisher, year, edition, pages
2004. 165-175 p.
Collision avoidance, Statistical decision making, Autonomus braking
Engineering and Technology Control Engineering
IdentifiersURN: urn:nbn:se:liu:diva-89176OAI: oai:DiVA.org:liu-89176DiVA: diva2:612722
8th International Forum on Advanced Microsystems for Automotive Applications, Berlin, Germany, March, 2004