Hybrid neural-physical modeling and enhanced multi-objective motion planning with Neuro-Fuzzy Adaptive Sliding Mode Control for dual-arm space robots

The dynamic coupling between the robotic arm and the base of a free-floating space robotic system presents significant challenges in modeling, planning, and achieving motion accuracy. Existing approaches are limited in their ability to address the system’s inherent dynamic nonlinearities. This paper proposes a hybrid model that retains the generalized-Jacobian analytical inertia matrix and uses a multilayer perceptron (MLP) to approximate the velocity-dependent generalized-force term of a dual-arm space robot. Furthermore, an Enhanced Hybrid Multi-Objective Optimization (EHMO) algorithm is proposed for trajectory planning. This framework integrates Multi-Objective Particle Swarm Optimization (MOPSO) with the Non-dominated Sorting Genetic Algorithm II (NSGA-II) through a novel parallel competitive mechanism and employs a Reference Vector Guided Evolutionary Algorithm (RVEA) for selection. To address control stability in the presence of uncertainties and external disturbances, a Neuro-Fuzzy Adaptive Sliding Mode Control (NFASMC) strategy is then introduced. NFASMC uses structured fuzzy initialization for its RBF gain scheduler, while online gain adaptation and a continuous fuzzy reaching mechanism target robustness and chattering suppression. Simulation results demonstrate the effectiveness of the proposed framework in improving modeling accuracy, trajectory-planning performance, and tracking accuracy for free-floating dual-arm space robots.