Dynamic path-speed planning algorithm for autonomous driving on structured roads

Aiming at the problem of local path planning for structured roads, this paper proposes a framework of local path-speed planning for autonomous driving, which takes safety as the premise and improves driving efficiency. The framework simulates human driving thinking and divides the local path planning of autonomous driving into two parts: lane decision and path-speed planning. In the part of the lane decision, a lane decision algorithm based on driving risk field and safe distance is proposed, which can ensure driving efficient and ensure that the planning vehicle is always in a low-risk driving environment. In the part of the lane change path-speed planning, a candidate path generation algorithm based on uniform sampling of lane change time and a cost function considering lane change timeliness, driving safety, speed smoothness, and path continuity are proposed to achieve optimal path selection and speed planning. In the experiment part, there are six different driving tasks. In six scenes, the local path-speed planning framework proposed in this paper can plan a safe, efficient, and smooth driving path and a safe planning speed. Taking the scenario of detouring low-speed obstacles as an example, the path-speed planning algorithm proposed is compared with the path-speed planning algorithm based on discrete optimization in Hu et al. It has been verified that the algorithm proposed can ensure that planner is always at low environmental risks and drive with high driving efficiency.