Aiming at the problems of rapidly exploring trees(RRT)in some complex environments,such as low search efficiency,slow convergence rate and many redundant nodes,an improved RRT algorithm is proposed.Firstly,the adaptive target probability strategy is introduced to adjust the sampling probability of the target point in real time. Secondly,node turning strategy is introduced to improve the success rate of single sampling.Finally,redundant nodes are clipped to the generated path to make the path more in line with the actual application requirements.Simulation experiments are carried out in MATLAB and compared with RRT algorithm and RRTGoalBias algorithm. The experimental results show that the improved algorithm has good adaptability in a variety of different environments,and the running time,sampling times and sampling success rate are greatly improved.The final average path length is reduced by 21.1%,and the average node number is reduced by 75.3%,which proves the superiority and practicability of the improved algorithm.