Abstract: With the development of computer and other electronic technologies, Autonomous Underwater Vehicle (AUV) have shown great potential in the field of marine scientific research and other underwater engineering, assisting engineering personnel in completing various underwater tasks. In most ocean exploration scenarios, AUVs need to autonomously plan routes and navigate according to the specified routes. Traditional path planning methods using intelligent optimization algorithms such as cuckoo search algorithm have issues such as unreachable search targets, slow convergence speed, and poor obstacle avoidance capabilities. Although the A* algorithm based on grid maps solves these problems, it fails to meet the smoothness requirements of AUV routes. Therefore, this study integrates the A* algorithm with Bézier curves to improve path planning, focusing on maintaining AUV underwater posture. The improved algorithm is validated through simulation experiments using a simulation environment built with Python on the Windows 11 operating system. The results show that the AUV path planning based on the A* algorithm and Bézier curves can not only make the path shortest, but also further smooth the entire path while ensuring safe navigation, which is beneficial for maintaining the stable attitude of the AUV during operation.