Adaptive bézier curve-based path following control for autonomous driving robots

Abstract

This article presents a concise and efficient path-following strategy, along with a set of real robot experiments to evaluate its superior performance. The following trajectory is generated in the form of a quartic Bézier curve with an adaptive control point generation method based on the integral length and curvature of the reference path. An impressive merit is that the cutting-corner problem during sharp turns can be avoided and smooth speed regulation can be achieved automatically. Another advantage is that the robot can quickly return to the reference path from a large lateral position or heading deviation, without any large space requirement for adjustment. The first few commands derived from the differentiation of the following trajectory are utilized. Simulation results show that the proposed method has a higher accuracy under the same-level computation time compared with other simple geometric methods. Real-world robot experiments are conducted in various environments to verify the proposed algorithm's accuracy, robustness, and flexibility. The average path-following error of real-world experiments is under 0.1 m, even with sudden path changing for obstacle avoidance. Additionally, with the proposed algorithm, the robot can navigate safely in a residential community where frequent pedestrian incursions occur.