Curve warning driver support systems. A sensitivity analysis to errors in the estimation of car velocity

Abstract

Past research projects on intelligent vehicles have already led to the development of a large number of Advanced Driver Assistance Systems. The current research focus is now shifting towards integration and adaptive automation systems that share the control between driver and the machine. Artificial co-drivers can be used for this scope, as tutors to provide holistic support to the driver. However the question of the accuracy and robustness of co-driver evaluations, with respect to perception noise, becomes critical. This work discusses the robustness to perception noise of a Curve Support function, as part of a holistic driver support system based on a co-driver concept. The main objective of the work is to respond to the following question: how accurate should the vehicle state estimation be to design a reliable system? The paper gives a general framework and preliminary results related to noise in the estimation of the vehicle velocity vector.