Active Momentum Control for Spinal In-Pipe Robots in Microgravity: On-Orbit Validation

Locomotion in confined microgravity environments, such as spacecraft piping, presents unique challenges due to the lack of gravitational normal forces and the strict requirement for environmental safety. Traditional rigid-body controllers often struggle to generate sufficient traction for agile maneuvering without violating the structural limits of fragile surroundings. This article introduces a novel control framework for a 6-link spinal wheeled-legged robot that actively exploits internal spinal dynamics to steer the system. We propose a lie-algebraic error-state model predictive control that plans optimal momentum fluxes and spinal configurations, ensuring singularity-free tracking even during omnidirectional tumbling. To guarantee safety, a hierarchical whole-body controller enforces strict environment-admissible force constraints and integrates an event-triggered asymmetric friction adaptation strategy to handle uncertain surface properties. Uniquely, the proposed framework is validated through real-world on-orbit experiments using a flight-equivalent pipe testbed aboard the China Space Station. Results demonstrate that our method successfully achieves agile heading recovery where rigid-body baselines fail to converge due to traction loss, while strictly maintaining contact forces within the 6.6 N safety limit despite friction uncertainties.