In view of the complexity of the marine environment, considering the limitations of the AUV fuel and other factors, in order to reduce the energy consumption of the AUV in the unknown environment, taking an AUV as the research object, a path planning algorithm of RRT is designed and implemented. An AUV path planning algorithm based on improved RRT (Rapidly-exploring random tree, RRT) and Bezier optimization is proposed to accelerate the search, avoid tortuous paths and improve path smoothness. In this method, the cost formula of algorithm is used as the sampling standard, and the pruning strategy is used to simplify the path. Then, Bezier curve fitting is used to generate the initial solution of the planning problem, so as to improve the feasibility of generating the path. In addition, local re-planning is used to solve the obstacle avoidance problem in the unknown environment. In the simulation experiment, the algorithm is compared with the conventional RRT method, and the results show that the algorithm has a great improvement in search speed, smoothness, feasibility and other aspects.