Dynamic modeling and learning based path tracking control for ROV-based deep-sea mining vehicle

Track slippage and body sinking of the tracked mining vehicle in the traditional deep-sea mining system are the critical issues for operating stability. To solve this bottleneck problem, a novel ROV-based deep-sea mining system is proposed in this study, in which a remotely-operated vehicle (ROV) towering a sledge-shaped mining robot (MRT) named ROV-based Deep-sea Mining Vehicle (ROVDMV) is instead of the traditional tracked Deepsea mining vehicle. The design of the ROVDMV can fundamentally overcome the bottleneck problem. However, the complex marine environment and multi-rigid-body design of the ROVDMV pose new challenges for its pathtracking control. Firstly, the dynamic model of the ROVDMV considering the ROV at a fixed depth is established based on the bicycle model, which is mainly used as the control object in the numerical simulation. Secondly, a learning-based path-tracking control strategy is proposed for the path-tracking control of the ROVDMV. In the control strategy, a novel nonparametric learning (NPL) method is introduced to learn the uncertain nonlinear dynamics considering the external disturbances and parametric uncertainty. The NPL method is proven to provide bounded estimated error. Besides, the enhanced NPL method can save approximately 33 % of the computation time, and the average computation time for its optimization control problem is only 12.47 ms. Finally, the numerical results show that the NPL method can learn nonlinear dynamics accurately, and the proposed strategy has proven to be effective.