Research On Collision Avoidance Path Planning Of Car-like Robots

This thesis focuses on the research of collision avoidance path planning for car-like robots. The main research topics and contributions are as follows:At the beginning of this thesis related issues of collision avoidance path planning are introduced. Configuration space, environment presentation technologies, planning methods, search algorithms are reviewed, which must be solved and are covered by the scope of path planning.Based on the analysis of the special nature of path planning for car-like robots, a new algorithm named two level planning method is proposed. Considered the vehicle movement nature of car-like robots, the arc or line local path planning is used On the first level planning, which is the basis of the second level planning in which improved quick dijkstra algorithm and improved wave expansion algorithm are used. The differences between this method and the previous methods are as follows: It takes into account the motion property of car-like robots, so in the path search every step is a practical path of car-like robots. It is not cell for the basic unit of searching, but a section of short path along which car-like robots can actually walk. In this way, the scale of the cell can be as small as possible to improve accuracy and ensure the accuracy of the environment presentation, also in the search algorithm many unessential cell nodes can be saved, which improves the time efficiency. Meanwhile, improved dijkstra algorithm overcomes the deficiency of the space consumption and its search speed is improved. Improved wave expansion algorithm overcomes the shortcomings of the space consumption and the deficiency that the path which wave expansion algorithm finds is not overall optimal. Theory and the examples show that the new algorithm ideas are novel and the conclusions are right.Finally, after the issues of path planning for multiple car-like robots are in-depth studied, multi-level conflict resolution method is proposed. This work is of great significance for both making the content of path planning for multiple car-like robots rich and the practical application.