Research On Optimization Strategy Of RRT Algorithm For Dynamic Environment

Robots can assist or replace humans in complex or dangerous operations environments to complete tasks,which effectively ensures the safety of human life and enhances certain economic benefits.The path planning problem refers to finding a collision-free path connecting a starting point to a target point in given workspace,thus providing technical support for mobile robots to perform tasks.The Rapidly-exploring Random Tree(RRT)algorithm is an intelligent path planning algorithm based on sampling.It finds feasible path in the workspace quickly without modelling the environment and with low computational complexity.But,the problems of searching for path in dynamic environments include high randomness,the slow convergence speed and numerous inflection points of path.Therefore,this thesis takes mobile robots as the research object to investigate the problem of optimization strategy of RRT algorithm in dynamic environment.The main research contents are as follows:(1)A fast path planning algorithm(Fast Rapidly-exploring Random Tree star,FastRRT~*)is proposed based on a hybrid sampling and backtracking wiring strategy to solve the problem of low efficiency of the RRT algorithm in searching for path in complex static environments.The randomness of sampling and the overhead of searching path in the RRT algorithm are reduced by the hybrid sampling and backtracking wiring strategy,and the path are smoothed based on the cubic B-sample algorithm to improve the feasibility of the path.Finally,the effectiveness and superiority of the proposed algorithm are verified through numerical simulations and virtual experiments.(2)A High-quality Dynamic Rapidly-exploring Random Tree star(HQD-RRT~*)is proposed based on the improved RRT~*-SMART algorithm and local obstacle avoidance strategy to address the problem of low navigation capability of the RRT algorithm in dynamic environments.A high-quality initial path is planned based on the improved RRT~*-SMART algorithm and the node information of the tree structure is stored as a priori knowledge.During navigation along the initial path,local obstacle avoidance is performed based on the priori knowledge and the data information acquired by sensors,improving the effectiveness of the RRT algorithm for obstacle avoidance in dynamic environments.(3)A dynamic path planning method based on the Back Tracking Rapidly-exploring Random Trees star(BT-RRT~*)algorithm and adaptive local obstacle avoidance strategy is proposed for the collaborative path planning of multiple mobile robots in dynamic environments.Firstly,the time cost of RRT path search is reduced by the BT-RRT~*algorithm to ensure that each mobile robot has a better initial path.Then,the collaborative obstacle avoidance between multiple mobile robots is realized according to the adaptive local obstacle avoidance strategy.Finally,the effectiveness of the algorithm is verified by simulation experiments,and the experimental results show that the method improves the navigation capabilities of multi-mobile robots in a dynamic environment.