Robust Safety-Critical Control of Planar Robotic Manipulators Under Disturbances

Abstract

We present a safety-critical control framework based on Disturbance Observer-Parameterized Control Barrier Functions (DOP-CBF) for general n-DOF planar robotic manipulators. The framework enforces joint angle and joint angular velocity limits as key safety constraints. It integrates a proportional-derivative (PD)-like nominal controller, a momentum-based disturbance observer, and an adaptive safety filter in a unified design. Using online estimates of lumped disturbances and their cross-joint effects, the method adaptively adjusts safety margins in real time, removing the need for prior knowledge of uncertainty bounds while accounting for disturbance-induced coupling effects commonly encountered in robotic manipulators. Our theoretical analysis establishes Input-to-State Stability of the closed-loop system. The proposed approach is validated through simulations on a 2-DOF planar manipulator under multiple scenarios. Compared with an Explicit Reference Governor (ERG), it suppresses transient overshoot and oscillations in disturbance-free conditions and maintains robust performance under Gaussian white noise, while mitigating the severe oscillations observed in the baseline.