A local trajectory planning and control method for autonomous vehicles based on the RRT algorithm

This paper presents a local trajectory planning and control method based on the Rapidly-exploring Random Tree algorithm for autonomous racing vehicles. The paper aims to provide an algorithm allowing to compute the planned trajectory in an unknown environment, structured with non-crossable obstacles, such as traffic cones. The investigated method exploits a perception pipeline to sense the surrounding environment by means of a LIDAR-based sensor and a high-performance Graphic Processing Unit. The considered vehicle is a four-wheel drive electric racing prototype, which is modeled as a 3 Degree-of-Freedom bicycle model. A Stanley controller for both lateral and longitudinal vehicle dynamics is designed to perform the path tracking task. The performance of the proposed method is evaluated in simulation using real data recorded by on-board perception sensors. The algorithm can successfully compute a feasible trajectory in different driving scenarios.