An Obstacle Avoidance Method for Autonomous Vehicle in Straight Road Based on Expanded Circle

In this paper, an algorithm is proposed to avoid online obstacle in a straight road or in an unknown environment. The algorithm is built upon the concept of expanded circle, which can be used to calculate the required steering angle for the vehicle. And then a lateral controller is designed to maintain the desired steering angle. Moreover, to keep the same posture of the vehicle after the obstacle avoidance manipulation, a constraint of the terminal point pose is added. The algorithm is computationally efficient, and guarantees online obstacle avoidance. Its performance is validated by actual experiments on autonomous vehicles in a laboratory setting and the results show that the proposed algorithm is more efficient than DWA, and even better than real human drivers.