Anti-Disturbance Attitude Control of Space-Flying Robot Based on Neural Dynamic Inverse

This study investigates the attitude control problem for space-flying robot (SFR) that works inside the space station. A neural nonlinear dynamic inverse (NDI) is proposed to handle possible dynamic couplings among roll, pitch and yaw channels of SFR, where a neural network is carefully constructed to reduce the computing complexity of traditional NDI algorithm. Based on the decoupled attitude dynamics, a low-complexity disturbance observer-based attitude control is designed for each channel, which can remove the negative effect of mismatched lumped disturbances from the output channel. The effectiveness of the proposed neural NDI-based anti-disturbance control is verified by both simulation and experiment for a real SFR.