Fault-tolerant control and safe shutdown of free swing joints of space manipulator based on trajectory optimization

Abstract

Space manipulators are prone to free-swinging joint failures during on-orbit servicing missions, which turn the system from fully actuated to underactuated. Conventional emergency braking strategies result in slow energy dissipation, large base attitude disturbances, and high collision risks, making safe and smooth shutdown difficult to achieve. This paper proposes a Trajectory Optimization-based Safe Stopping (TO-SST) method. First, the system with a free-swinging faulty joint is modelled as an underactuated system, where the faulty joint is treated as a passive degree of freedom driven only by dynamic coupling. The torques of the healthy joints are used as control inputs. An optimal control problem is formulated with the objective of minimising kinetic energy dissipation, base attitude deviation, and control effort, while satisfying joint limits, velocity bounds, obstacle-avoidance constraints, and a terminal zero-velocity condition. The Gaussian pseudo-spectral method (GPM) is then employed to globally optimise the motion trajectories of the healthy joints, exploiting the dynamic coupling between active and faulty joints to guide the faulty joint to a smooth stop. Simulation tests on a planar 3-DOF space manipulator show that, compared with traditional Emergency Braking (EB), Model Predictive Control (MPC), and Adaptive Sliding Mode Control (ASMC), the proposed TO-SST method achieves a 100% success rate over 100 random initial conditions, with an average stopping time of only 4.2 s, a maximum base disturbance within 3.5°, and zero constraint violations. In terms of solver efficiency, GPM has an average solution time of 2.3 s, and its objective function value is 12% lower than that of Sequential Quadratic Programming (SQP) and 18% lower than that of Particle Swarm Optimization (PSO). The results demonstrate that TO-SST effectively harnesses the dynamic coupling between healthy and faulty joints, rapidly dissipating energy while maintaining base stability and strictly satisfying all safety constraints, outperforming the compared strategies. The method offers a practical and efficient solution for emergency response to on-orbit manipulator failures. Future work will extend the approach to high-DOF spatial manipulators and address modelling uncertainties, sensor noise, and multiple concurrent faults.