Mobile Robot Path Planning Based On Sampling Algorithm

Path planning is the key technology for autonomous navigation of mobile robots.The task of mobile robot path planning is to find a collision-free path from the initial position to the target position in the working environment,given the starting position and the target position.Aiming at some key problems of sampling based algorithm in mobile robot path planning,this paper proposes corresponding improvement strategies.The main work is as follows:1.Aiming at the problem of unreasonable distribution of milestones in narrow channel path planning based on PRM algorithm,a hybrid sampling method based on convex recognition is proposed.Through local sampling test,milestones are densely allocated at the convex edge of obstacles,and a small number of milestones are allocated in the open area,so that the global distribution of milestones is reasonable,so as to improve the efficiency of path search.2.In order to solve the problems of unreasonable distance metric,slow convergence speed and tortuous path in RRT algorithm in mobile robot path planning,two typical mobile robots,vehicle and differential robot,are taken as research objects,and corresponding improvement methods are proposed.For vehicles,a RRT algorithm based on distance metric learning is proposed.The shortest path curve is introduced as distance metric,and the supervised model is used to learn the metric offline to speed up online real-time computation.For differential robots,an improved RRT algorithm is proposed.Distance metric considers the steering constraint of differential robots,expands by dynamical step size to speed up convergence,prunes and smoothes the path,and finally obtains a smooth path that is easy to implement.3.In order to improve the real-time planning efficiency of path replanning based on RRT algorithm,a new method of path replanning based on improved bidirectional RRT algorithm is proposed.The random tree obtained by RRT algorithm is stored offline.The random tree is loaded at each replanning process,and two separate random trees are obtained by deleting the nodes and edges coincidence with the threat body.Based on these two trees,the new path is generated by bidirectional RRT algorithm.The simulation environment is developed by C++,QT and Python.A large number of simulation experiments are designed and carried out in this simulation environment.The effectiveness and superiority of the proposed method is verified in terms of path search success rate,path quality,path search speed and so on.