Research On RRT* Algorithm Of Path Planning For Mobile Robot

As a hot research in the field of robot,path planning of mobile robot is attracting more and more scholars to participate in it.Among many path planning algorithms,the sampling-based algorithm,Rapidly-exploring Random Tree,and its related algorithm greatly reduces the computational amount of path planning in high-dimensional space because it does not need to model the whole space,thus effectively solving the path planning problem in high-dimensional space and under complex constraints.Therefore,this paper will focus on RRT algorithms and proposes two improved algorithms.Firstly,for the low efficiency of path planning and randomness of sampling in RRT-Connect algorithm,the DRRT-Connect algorithm is proposed.The main improvements of the algorithm include:1)The algorithm adds a third node in RRT-Connect,so that the improved algorithm can generate four random trees at the same time in the extension,and accelerate the exploration speed of the random trees to the state space.2)An adaptive step size regulation function is added,which increases the extension step size of the random tree when exploring the space without obstacles,so as to improve the extension speed of the random tree.3)The target bias strategy is added on the basis of RRT-Connect algorithm,so that the improved algorithm does not need random sampling in the barrier-free area,but automatically uses random sampling to explore the obstacle area,so as to accelerate the algorithm extension speed and avoid falling into the local optimal.Secondly,to solve the problem of slow convergence rate of RRT* and its improved algorithm B-RRT*,an EB-RRT* algorithm is proposed,which integrates B-RRT* algorithm and DRRT-Connect algorithm.The main improvements of the algorithm include:1)An intelligent sampling function is introduced to make the extension of the random tree more directional,so as to reduce pathfinding time and improve the smoothness of the path.2)On the basis of B-RRT* algorithm,EB-RRT* algorithm added a rapid extension strategy to make the improved algorithm rapidly extend in free space by using the extension method of RRT-Connect algorithm.While in the obstacle space,the improved RRT* algorithm was extended to improve the extension efficiency and avoid the algorithm falling into local optimization.Finally,typical algorithms such as RRT,RRT-Connect,RRT* and B-RRT* were used as comparisons to carry out simulation experiments on the improved algorithm respectively.The results showed that the improved algorithms were significantly better than original algorithms in terms of path planning efficiency,iteration times and path length.Compared with RRT-Connect algorithm,the improved DRRT-Connect increased the efficiency of path planning by 37.9% and reduced the number of iterations by 19.3%.Compared with B-RRT* algorithm,the improved EB-RRT* algorithm increased the efficiency of path planning by 73.0%,decreased the number of iterations by 46.0%,and reduced the path length by 9.5%.