As the environment of underwater is complex and uncertain, it is hard to found exact mathematical model to control autonomous underwater vehicles (AUVs). Fuzzy control is combined with line of sight in this paper for AUV. Based on fuzzy control of obstacle avoidance for bionic robot fish is proposed and this algorithm provide an important basis for the path planning for AUVs. Through expert knowledge and experience, a fuzzy controller is utilized to adjust the moving direction for an AUV in order to make the AUV avoid the obstacle and then reach the goal securely. Results show by simulation and experiment that this algorithm is effective and feasible for the path planning for AUVs.