A target tracking system using sensors of multiple modalities

Abstract

Radar sensors and camera based vision systems will be used to provide object data in many Advanced Driver Assistance Systems (ADAS). In this paper, a multi-sensor target tracking system that combines data from a Frequency-Modulated Continuous Wave (FMCW) radar, multiple Short-Range Radars (SRR), a camera-based object detection system and vehicular dynamic sensors is described. Each object sensor individually measures the range, range rate and azimuth angle information of all objects within the observation region. The proposed system 1 groups objects from different sensors in overlapped observation region; 2 tracks an object across different sensor field of views; 3 reports the Cartesian coordinates of objects with improved accuracy and reduced rates of false detection and missed detection. The proposed target tracking system was implemented in a retrofitted vehicle. Only about two-percent CPU usage is needed for an 800MHz embedded processor. The output data was directly used by several vehicle features such as low-speed collision avoidance and Adaptive Cruise Control (ACC) systems. Copyright © 2013 Inderscience Enterprises Ltd.