Optimal trajectory planning for flexible truss assembly based on nonlinear model predictive control

With the continuous advancement of space technology, on-orbit assembly of truss structures using robotic manipulators is expected to become an important development trend. Unlike rigid payloads, large-scale lightweight trusses exhibit significant elastic vibrations during assembly, which can compromise end-effector accuracy and increase the risk of structural fatigue if they are not adequately suppressed. This paper proposes an optimal vibration-suppression trajectory-planning method for robotic manipulators assembling flexible truss structures. First, a rigid–flexible coupled dynamic model of the robot–truss system is established, and the interaction between manipulator end-effector motion and truss vibration is derived in task space. On this basis, a nonlinear model predictive control (NMPC) framework is formulated to optimize the assembly trajectory subject to vibration-suppression objectives, actuator limits, workspace constraints, and obstacle-avoidance requirements. By jointly penalizing terminal-state errors, structural modal responses, and control-input variations, the proposed method reduces modal excitation during trajectory generation. In the numerical study, the NMPC problem is solved sequentially in a receding-horizon manner using the current simulated state; however, no hard real-time computation deadline is imposed. In the physical experiments, the trajectories are generated offline and subsequently executed by the manipulator’s low-level position controller. The simulation and experimental results show that the proposed approach reduces multi-modal vibration, generates smooth and collision-free end-effector motion, and outperforms the considered non-optimized trajectory-planning baselines.