Abstract
A local path planning algorithm for intelligent vehicles under structured road environments is proposed based on collision risk evaluation. Firstly, the cubic spline curve is used to express the global path. Secondly, the candidate paths are generated considering vehicle safety and path smoothness, and the generated candidate paths are evaluated by the normalized cost function. Finally, the optimal local path is obtained at the current state under the guidance of global path information. For the cost function, the driving risk field is proposed, and a static and moving obstacle risk model is designed, where the safety cost function is established based on collision risk evaluation combined with the Gaussian convolution. Concerning with the curvature and path-following effect of the generated path, a path smoothness and an offset cost function are respectively built. The simulation results show that the proposed algorithm can generate smooth, collision-free paths in real-time, and the test intelligent vehicle can effectively avoid static and moving obstacles under various road scenarios.
| Translated title of the contribution | Local Path Planning for Intelligent Vehicles Based on Collision Risk Evaluation |
|---|---|
| Original language | Chinese (Traditional) |
| Pages (from-to) | 28-41 |
| Number of pages | 14 |
| Journal | Jixie Gongcheng Xuebao/Chinese Journal of Mechanical Engineering |
| Volume | 57 |
| Issue number | 10 |
| DOIs | |
| Publication status | Published - 20 May 2021 |