An adaptive data-driven controller for underwater manipulators with variable payload

The underwater environment poses a complex problem for developing control systems manipulators. Modeling the system is complicated and costly process due to highly nonlinear dynamics presence of unknown hydrodynamical effects. Furthermore, manipulators are usually deployed on Autonomous Underwater Vehicles (AUVs) which further influence change arm. This aggravated in operations where manipulating different objects necessary. These diverse manipulation tasks introduce external disturbances can lead fast degradation performance. In this article, we propose novel adaptive controller robot that have handle varying payloads with masses, geometries, buoyant forces. proposed strategy utilizes data-driven model an optimal formulation based neural networks. Moreover, developed online tuning strategy, interaction theory, allows gains be updated respect set performance metrics. Experiments were performed robotic arm variety while mounted both fixed base free-floating vehicle. We present number simulated experimental results illustrate benefits strategy. addition, comparative study against classical Model Predictive Control (MPC) demonstrates our proposal.