Credibilistic multi-sensor fusion for real time application.: Application to obstacle-detection and tracking

In. the road situation, one sensor provides some information which are imperfect (inaccurate and uncertain) on the perceived targets. This sensor can also be unreliable. Experience from previous researches and development programmes shows, that individual sensorial components are not sufficiently reliably in complex traffic scenarios. Fixed target detection and handling complex traffic configurations in urban areas need improved perception systems and methods. Given the output of different sensors and image processing modules, the purpose of the data fusion in the European CARSENSE project is to build the most reliable, accurate and complete map of the vehicle environment featuring the neighboring obstacles. Such a map is of primary interest in the design of a robust and reliable driver assistance system. Moreover, the fusion problem addressed in CARSENSE is basically a multi-target tracking problem. In this paper, we present the credibilistic approach for multi-sensors data association. This association algorithm provides a reliable and robust representation of the environment using all available information.