Controllable Clothoid Path Generation for Autonomous Vehicles
Songyi Zhang, Runsheng Wang, Zhiqiang Jian, Wei Zhan, Nanning Zheng, Masayoshi Tomizuka
Abstract
This letter proposes a novel and simple smooth path generation algorithm for autonomous vehicles. The proposed method can rapidly generate feasible and curvature continuous paths connecting any two given states with null curvature. The generated path comprises straight lines, circular arcs and clothoid segments that are close to human driving behavior. Moreover, while keeping an accurate connection with the boundary states, the proposed method also allows fine-tuning the curve with the maximum curvature of the path and the midpoint’s position, which makes it more flexible in avoiding obstacles while keeping feasibility. Various experiments show that the proposed method has extremely low time consumption and high robustness for different conditions. Meanwhile, it is more flexible than the recent research and the spline-based methods.