Q-learning Based Obstacle Avoidance Control of Autonomous Underwater Vehicle with Binocular Vision

Abstract

Autonomous Underwater Vehicle (AUV) is widely used in various marine tasks, in which the autonomous obstacle avoidance function is the key technology to achieve autonomous motion such as path planning and trajectory tracking. Autonomous obstacle avoidance requires AUV to effectively detect obstacle information and select the optimal control strategy through autonomous judgment to achieve the obstacle avoidance function. However the ocean environment is very complex, full of uncertainty and danger, the underwater information obtained by using acoustic detection technology is very limited, which makes the control of underwater robot very difficult. To address this problem, this paper proposes a method for obstacle avoidance by binocular visual parallax control (BVPC) technique. In addition, the motion control of the robot is realized by Q learning (QL). In particular, the control strategy can be applied to different underwater unknown scenarios instead of a specific one, improving the wide applicability of this control method. Finally, the effectiveness of the proposed scheme is verified by simulation.