Hybrid Neural Control of Electrically Driven Reconfigurable Manipulators Under Time‐Varying Constraints

ABSTRACT This paper presents a novel hybrid control strategy for force/position tracking in rigid‐link electrically driven reconfigurable manipulators (RL‐EDRM), explicitly considering actuator dynamics and time‐varying environmental constraints—a combination not previously addressed in the literature. Existing force/position control methods often neglect actuator nonlinearities and dynamic constraint variations, limiting their applicability in real‐world scenarios. In this work, the term “hybrid” refers to the integration of both model‐based and model‐free control schemes within a unified framework. Specifically, the controller combines a model‐based backstepping design, which manages the known manipulator dynamics, with a model‐free radial basis function (RBF) neural network that approximates unknown dynamics and unmodeled nonlinearities. Additionally, an adaptive compensator addresses external disturbances, friction, and approximation errors. The control scheme ensures asymptotic convergence of tracking errors and enforces constraint compliance by dynamically adjusting force and motion commands. Direct current (DC) motor models are incorporated to generate accurate current and torque profiles. System stability is analytically guaranteed using Lyapunov theory. Simulation studies on a 2‐DOF reconfigurable manipulator under time‐varying environmental conditions demonstrate the effectiveness of the proposed method. Quantitative results confirm the controller's robustness, adaptability, and enhanced performance compared to conventional model‐based strategies.