Optimised Multi-Arm Robot Motion to Reduce Satellite Disturbances During In-Orbit Assembly

Abstract

​​The increasing demand for large-scale space structures, such as modular telescopes and solar power stations, requires efficient and reliable In-Orbit Assembly (ISA) operations. Robotic systems play a central role in this process, but their locomotion can induce significant disturbances to the hosting spacecraft because of momentum conservation. These need to be actively compensated, at the expense of scarce and valuable fuel resources. This thesis investigates passive disturbance minimisation through optimised path planning for Cranfield’s Multi-Arm Robot for In-Orbit Operations (MARIO). A literature review identified key approaches to reaction-minimising motion planning, including real-time control, trajectory optimisation, and sampling-based planners. Based on this analysis, the Stochastic Trajectory Optimization for Motion Planning (STOMP) algorithm was implemented within ROS2 and MoveIt2, with a custom cost function penalising joint torques as an indicator of disturbance. The methodology was tested through high-fidelity Gazebo simulations of MARIO performing crawling locomotion on modular structures designed for laboratory experiments at Cranfield University’s ASTRA-Lab. Results demonstrated that mass distribution strongly influences disturbance magnitude, and that optimised trajectories can reduce disturbances by nearly 50% compared to linear baselines. However, issues such as inverse kinematics solver variability, limited repeatability, and collision risks limited practical applicability. The most promising results were obtained when combining Cartesian initial trajectories with STOMP optimisation, although the custom disturbance cost function proved less effective than anticipated. The findings highlight both the potential and challenges of trajectory optimisation for space robotics, offering insights into the coupling between manipulator motion and spacecraft dynamics. Future work should refine disturbance modelling cost functions, integrate hybrid planning approaches such as RRT* with STOMP, and extend validation to physical experiments with MARIO in the ASTRA-Lab. Overall, this research contributes to safer and more fuel-efficient in-orbit assembly operations, supporting the development of sustainable space infrastructure.​