A novel adaptive impedance control algorithm for underwater manipulators

Introduction Underwater manipulators performing precision operations are simultaneously affected by nonlinear hydrodynamic uncertainties and abrupt changes in environmental contact conditions, making it difficult for conventional fixed-parameter impedance control to maintain both trajectory-tracking accuracy and compliant interaction. To address this problem, this paper proposes an adaptive impedance control framework for underwater manipulators. Methods A multi-degree-of-freedom dynamic model is first established by explicitly considering rigid-body dynamics, added-mass effects, nonlinear hydrodynamic damping, and lumped uncertainties. The lumped disturbance is then decomposed according to its functional dependence into a state-dependent unmodeled hydrodynamic component and a residual exogenous disturbance component. An RBF neural network is employed to approximate the state-dependent nonlinear component, while a nonlinear extended state observer estimates the remaining online disturbance, thereby forming a composite feedforward compensation mechanism. On this basis, a prescribed-performance-based adaptive impedance law is developed to regulate the virtual stiffness and damping according to the tracking and contact-force errors while constraining transient performance. Lyapunov analysis establishes the uniform ultimate boundedness of the resulting closed-loop signals. A MATLAB/Simulink–ROS/Gazebo co-simulation platform is constructed to evaluate the proposed controller under time-varying current disturbances and abrupt contact-stiffness transitions. Results Repeated simulation trials under multiple disturbance conditions are conducted to evaluate the statistical robustness of the performance indices. Compared with traditional fixed-parameter impedance control, the proposed method reduces the end-effector trajectory RMSE by 82.1%, the peak contact force by 57.9%, and the force-settling time by 49.4%. Discussion These results demonstrate improved disturbance rejection and compliant interaction within the tested operating conditions, validating the effectiveness of the proposed architecture without requiring prior knowledge of environmental stiffness.