Path Planning for Autonomous Vehicle Based on a Two-Layered Planning Model in Complex Environment

الملخص EN

The autonomous vehicle consists of perception, decision-making, and control system.

The study of path planning method has always been a core and difficult problem, especially in complex environment, due to the effect of dynamic environment, the safety, smoothness, and real-time requirement, and the nonholonomic constraints of vehicle.

To address the problem of travelling in complex environments which consists of lots of obstacles, a two-layered path planning model is presented in this paper.

This method includes a high-level model that produces a rough path and a low-level model that provides precise navigation.

In the high-level model, the improved Bidirectional Rapidly-exploring Random Tree (Bi-RRT) based on the steering constraint is used to generate an obstacle-free path while satisfying the nonholonomic constraints of vehicle.

In low-level model, a Vector Field Histogram- (VFH-) guided polynomial planning algorithm in Frenet coordinates is introduced.

Based on the result of VFH, the aim point chosen from improved Bi-RRT path is moved to the most suitable location on the basis of evaluation function.

By applying quintic polynomial in Frenet coordinates, a real-time local path that is safe and smooth is generated based on the improved Bi-RRT path.

To verify the effectiveness of the proposed planning model, the real autonomous vehicle has been placed in several driving scenarios with different amounts of obstacles.

The two-layered real-time planning model produces flexible, smooth, and safe paths that enable the vehicle to travel in complex environment.