Intelligent vehicle local planning based on optimized path generation and selection

Local planning is of great importance to intelligent vehicle driving at high speed where the traffic environment is complex. The local planning must satisfy the performance of good real-time and guarantee feasibility and safety. In this paper, a novel local planning method based on the optimization of curve generation and selection is proposed. By optimizing the curve parameter model, we can get the feasible path which is satisfied with vehicle dynamics constraints. Then taking into account the obstacle constraint, we apply this method to generate a cluster of candidate paths used for tracking global path. Next, we use optimization indicators presented in this paper to choose the suitable desired local path. Experimental results show that: using our method, the change in the curvature of the generated desired local path is slight and it meet the requirement of safety and control smoothness. Besides, this method can satisfy the meeting of tracking global path. What's more, in real traffic environment, this method reflects a good performance in real-time and safety.