Robot Path Planning Algorithm For Complex Unknown Environment

Robot path planning is a key issue in robot navigation and localization system. A flexible and effective path planning algorithm will help Robot identify the environment automatically, and finally reach the destination to complete the assigned task.The following is the author's main work based on the existing methods.Firstly, an online path planning method is proposed. Based on the original rolling optimization method, the rule for sub goal mapping is modified; then we introduce the diagonal distance to A* search algorithm for local path finding, so as to better adapt the real time requirement, also a taboo search and diversification strategy is included to overcome the limitation of oscillation and dead circulation, which is always encountered in traditional rolling method. The flexibility and effectiveness is illustrated in simulation tests.Secondly, path planning in high dimension space with narrow passages is studied. The improved RRT (Rapid-exploring Random Tree) method is proposed. Based on the original RRT algorithm, the nearest neighbor search module is approximated by "diagonal distance" to reduce the computational burden, the idea of local searching trees is introduced for exploring the "hard reach" regions, ensuring the effective path can be found. Finally we add a module of path smoothing to make the robot work more effective. Simulation results prove that the algorithm effective and feasible.