RBF neural network

Adaptive impedance control enables precision underwater manipulation amid uncertainty

Precision underwater manipulation has long been compromised by unpredictable hydrodynamic forces and sudden contact changes. This paper tackles that challenge head-on. By pairing an RBF neural network with a nonlinear extended state observer, the proposed adaptive impedance framework achieves an 82.1% reduction in trajectory error and a 57.9% drop in peak contact force compared to conventional fixed-parameter methods. These are measurable, validated gains. For a broader perspective on adaptive control in dynamic marine environments, see our coverage of the hybrid glider path-planning framework.