GIB-RRT: a gamma interpolation bidirectional RRT algorithm

In a complex environment, when B-RRT and RRT* algorithms are used for path planning, there will be problems such as long planned paths, large number of iterations, low sampling efficiency and long search time. To solve these problems, this paper proposes a gamma interpolation bidirectional RRT algorithm––GIB-RRT. First, the algorithm uses a bidirectional search strategy to expand two random trees simultaneously to speed up the convergence speed. In the expansion process, an adaptive goal biasing strategy is introduced to improve the sampling efficiency, and the probability of expansion to the respective target point is continuously changed according to the number of collision detection failures. After the initial path is obtained, a greedy pruning algorithm is used to simplify the path points and reduce the path cost. An optimisation method of Gamma interpolation is then devised for the simplified path and combined with cubic uniform B-spline curve to generate shorter and smoothly executable path. The proposed algorithm is compared with B-RRT*, IB-RRT* and B-RRT in different complex environments in simulation experiment, and the results show that the proposed algorithm has better search efficiency and is able to obtain optimal path in the least time and with the most stable efficiency.