Design of an Underwater Autonomous Inspection Robot

Abstract

This paper mainly designs and manufactures an underwater autonomous inspection robot (UAIR), including mechanical structure, circuit control and visual recognition. Using computer-aided modeling and design software, combined with the theoretical knowledge of fluid dynamics, the mechanical structure of UAIR is optimized to prolong the underwater working time. The circuit module includes the design of the MPU and the design of the control algorithm. Through the sensor to monitor the trajectory and posture of the robot under the water, the motion controller is designed by using the PID principle to realize the dynamic adjustment of the robot and achieve the purpose of autonomous operation. Based on Yolov5 model, an end-to-end image deblurring scheme is proposed, and multi-scale restoration strategy is used to improve the accuracy of robot object recognition. Finally, the UAIR can operate stably within 20 meters underwater, and the autonomous control and remote control modes can be switched flexibly, with a recognition accuracy of 96%.