Path planning of unmanned vehicles in narrow and long space based on improved RRT algorithm

A path planning algorithm based on improved rapidly-exploring random trees (RRT) is proposed for the path planning system of unmanned vehicles in narrow and long space, which solves the problems of large randomness and lack of safety of the traditional RRT algorithm. The algorithm improves the traditional RRT algorithm by adding adaptive target probability sampling strategy and dynamic step size strategy. At the same time, considering the dynamics constraints of driverless vehicles in the actual situation, the algorithm adds vehicle collision constraints and path angle constraints, and proposes a random turning strategy within the restricted area to solve the problem that the angle constraints will lead to the multiplication of iterations, and a path with higher safety is finally obtained. The performance of the proposed algorithm is compared with existing algorithms by computer simulation. Compared with the traditional RRT algorithm guided by artificial potential field in narrow and long space, the iteration times, planning time and path length of the proposed algorithm are reduced by 33.09%, 6.44% and 0.06%, and the planning ability of the proposed algorithm is improved in both simple environment and dense obstacle environment. The proposed algorithm has higher planning efficiency and fewer iteration .