This work is about planning the trajectory for autonoumous vehicles in the parking sceniro. The vehicle movement is based the reeds-shepp curve and the planning algorithm is the hybrid A* algorithm. And we show you two parking scenes, parallel parking and backward parking.

The reeds-shepp curve is the basic movement for bicycle model vehicles in the obstacle free environment. It's usually used in the movement unit in the motion planning problem.

I think the basic theory of the algorithm is simple for every one. What I want to focus on is the kd-tree and the priority queue in the algorithm implement. The kd-tree code is down load from leet code and something wrong were corrected and the basic inrange function is attached.

The algorithm is not efficient and the construction of the obstacle map both with the policy map consume most of the time. For the convience of debug, I saved this data as .pkl file.

The following is the animination