Jiawen Li, Xiang Cao, and Xueyou Huang
ROV, YOLO, target detection, target grasping
With the increasing prominence of unmanned systems in aquaculture and fisheries, this study proposes a visual perception-based remote operated vehicle (ROV) target grasping strategy to address complex underwater environments and high operational risks. The strategy consists of two parts: target detection and target grasping. Target detection is to use the visual sensor carried by ROV to perceive the underwater environment, obtain environmental images, and use deep neural network algorithms to detect targets in the images. The detected target is the premise of realising the target grasping, and the target grasping uses a fuzzy PID algorithm to control the robot arm carried by the ROV to grasp the detected target. Simulation and experiments show that this method can realise target detection and grasping under different water quality and has higher detection accuracy and speed. In practical engineering applications, this method meets the requirements of intelligent aquatic fishing in complex underwater environments.
Important Links:
Go Back
