Active Disturbance Rejection Control for Tilt-Rotor Aerial Manipulation Robots
Abstract
While fully actuated tilt-rotor aerial robots offer enhanced dexterity over conventional systems, their complex architectures are susceptible to multi-band disturbances and unmodeled actuator dynamics. This work presents a comprehensive control framework to address these challenges. First, a novel frequency-selective nonlinear disturbance observer is developed by combining a linear estimator with an adaptive super-twisting structure through frequency allocation, enabling effective estimation of composite slow and fast varying disturbances. Building on this observer, a nonsingular terminal sliding mode controller is designed to ensure singularity-free finite-time tracking. Furthermore, a model-reference-adaptive-control-based allocation scheme was proposed to compensate for unmodeled servo dynamics at the actuator level. Extensive simulations and real-world experiments, including inclined peg-in-hole, overhead grasping, outdoor flight, and sustained <inline-formula><tex-math notation="LaTeX">${\bf 90^\circ }$</tex-math></inline-formula> pitch maneuvers, validate the effectiveness of the proposed framework. Compared with state-of-the-art baselines, the proposed method reduces the attitude and velocity root-mean-square errors by up to 30.9% and 20.2%, respectively, while maintaining the attitude error within <inline-formula><tex-math notation="LaTeX">${\bf 3.9^\circ }$</tex-math></inline-formula> during <inline-formula><tex-math notation="LaTeX">${\bf 90^\circ }$</tex-math></inline-formula> pitch maneuvers.