Abstract
Road estimation and obstacle detection are two key techniques for autonomous driving vehicles, requiring high accuracy, efficiency and robustness. In this paper, we propose an efficient nonparametric road disparity estimation algorithm that makes full use of the characteristics of the road disparity. It can adapt to both planar and nonplanar road. In addition, to satisfy the real-time requirements of the application, we exploit a new texture gradient based constraint to compute the free space and detect the obstacle, which will also help to obtain more accurate results. Experimental results with real stereo sequences show that the proposed algorithm outperforms the standard methods both in terms of detection accuracy and runtime performance, which can run over 200 fps with normal PC.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call