Traditional Method Meets Deep Learning in an Adaptive Lane and Obstacle Detection System

Recently, a lot of researches and applications regarding to autonomous vehicle have been invested and developed. These applications not only reduce the risk of traffic accidents but also bring convenience to the driver. In these applications, lane and obstacle detection are indispensable tasks. Therefore, in this paper, an adaptice lane and obstacle system has been proposed to solve the above issues. In particular, our model not only detects the lanes but also avoids the obstacle on the road. Our system is tested on the simulated environment using Unity software then embedded on a 1:7 RC vehicle and tested on a small driving environment. The experimental results demonstrate the effectiveness and robustness of the proposed model in many challenging situations.