Multiple sensor collision avoidance system for automotive applications using an IMM approach for obstacle tracking

In this paper a multi-sensor collision avoidance system is presented for automotive applications. For obstacle detection and tracking, millimeter wave (MMW) radar and a far infrared (FIR) camera are chosen in order to provide object lists to the sensors' trackers respectively. The algorithm for track management, data association, filtering and prediction for both sensors is also presented, focusing on Kalman filtering. Thus, an interacting multiple model (IMM) filter is designed for coping with all possible modes in which a car may be. Finally, a distributed fusion architecture using a central track file for the objects' tracks is adopted and described analytically. The results of the work will be used, among others, in the European co-funded project "EUCLIDE".