Real-time lane detection and departure warning system on embedded platform

Abstract

Within the last few years, studies on Advanced Driver Assistance Systems (ADAS) have been actively conducted and deployed in modern vehicles; moreover, lane detection and departure warning systems are important modules of ADAS. However, most of the recent papers have only focused on PC-based lane detection modules, and very few concerns have been addressed for the customized embedded board. This paper proposes a real-time lane detection and departure warning technique on a commercial embedded board. The technique is based on Inverse Perspective Mapping (IPM) generating a topview image of the road and Kalman filter tracking removing noise and enhancing accuracy. The experimental results show good performance with an average correct detection rate of 96% under various challenging urban and highway conditions while the processing time takes only 22.76 ms per frame (1280×720) on the embedded board which verifies that the proposed method could be feasible for real-time applications in commercial ADAS products.