Detection-level fusion for multi-object perception in dense traffic environment

Due to much imperfect detection performance of onboard sensors in dense driving scenarios, the accurate and explicit perception of surrounding objects for Advanced Driver Assistance Systems and Autonomous Driving is challenging. This paper proposes a novel detection-level fusion approach for multi-object perception in dense traffic environment based on evidence theory. In order to remove uninterested targets and keep tracking important, we integrate four states of track life into a generic fusion framework to improve the performance of multi-object perception. The information of object type, position and velocity is made use of to reduce erroneous data association between tracks and detections. Several experiments in real dense traffic environment on highways and urban roads are conducted. The results verify the proposed fusion approach achieves low false and missing tracking.