Cooperative fusion for road obstacles detection using laser scanner and camera

In order to account for robustness of Automotive safety applications, fusion of data from multiple sensors is of remarkable importance to know the position of road obstacles. Challenges arise in Multi Sensor Data Fusion (MSDF) due to sensor uncertainty, multiple occluding targets and clutter by changing weather conditions. The proposed architecture address the problem by fusing information cooperatively from Laser scanner and monocular camera for robust detection of scene objects in the vehicle environment. The Fusion steps in the proposed method involve the application of the M-estimator SAmple Consensus (MSAC) algorithm for ground plane removal and density based clustering of laser data. Then the filtered laser objects are projected on the image plane and the corresponding region of interest (ROI) is extracted to localize the potential targets. Experimental results on challenging scene sequences of benchmark data sets prove the robustness of proposed fusion architecture for detecting vehicles on the road.