Sea trial experiment of AUV obstacle recognition and localization based on multi-beam sonar

During underwater navigation, the AUV may encounter unknown obstacles and needs to identify them to prevent collision and avoid damage. This paper This paper improves the accuracy of real-time calculations through binary filtering, connected component segmentation based on contours, and image temporal methods, thereby enabling the identification and localization of obstacles. Finally, the experiment is verified in the actual environment. The vision computer of the lower part of the vehicle can recognize and process the obstacle information, and the control computer can receive the obstacle information in real time to control the equipment to cooperate to avoid the obstacle. The research results of this paper provide input parameters for obstacle avoidance planning, and can provide reference for real-time obstacle avoidance operations of vehicles at sea.