Aiming at the complexity and uncertainty of working environments of underwater robot, in order to improve the capacity of interacting with the external environment and autonomous navigation for autonomous underwater robot, a local path planning strategies of underwater robot combined line-of-sight navigation principles with fuzzy control algorithm in an unknown environment was proposed, and real-time obstacle avoidance was achieved by the robot. Obstacles and target information within a certain range was obtained by using the ranging sonar and short baseline system to detect the environment. The deflection angle of the underwater robot was real time adjusted by adopting a fuzzy controller, and then the obstacles could be effectively avoided to reach the predetermined target point by the robot. Finally, a simulation experimental platform based on simulation by MATLAB was built. The number, sizes, shapes, and locations of obstacles were interactively set. The robot can avoid multiple obstacles in different road environments by adopting the fuzzy algorithm, which verifies effectiveness of the algorithm.