An Anti-disturbance Control Strategy for Underwater Peg-in-Hole Operation

In the proximity stage in the peg-in-hole operation, the control effects of underwater manipulator is strongly affected by external disturbance and model uncertainty, including ocean currents, friction, and imprecise model of underwater environments, all of which make it difficult to realize precise control in the task of the peg-in-hole. To this end, disturbance observer and radial basis function neural network are introduced to underwater manipulator control framework to solve the issue. Disturbance observer introduces equal compensation based on the observation results of model error from external disturbance estimation. Radial basis function neural network reduces imprecise parameters through online training, which achieves model uncertainty approximation. The peg-in-hole proximity experiment is carried out in an actual pool environment to verify the effectiveness of the motion control strategy proposed. Compared with other control strategies, the proposed control scheme reduces the root mean square error of the underwater manipulator’s motion by 52.91%.