Purpose Robots in space stations can perform point-to-point locomotion by utilizing reaction forces from the cabin walls, similar to human astronauts. However, body rotation and detachment from the cabin wall complicate control. This paper aims to develop a motion planning method that enables stable, accurate maneuvers under the unique constraints of zero gravity in a space station cabin. Design/methodology/approach An offline planning method based on an adaptive mass-spring model is proposed. The limb segment generating the reaction force is modeled as a virtual spring, and the desired maneuvering direction and velocity are adaptively mapped to the spring force direction and spring constant. Based on the mechanical design of the Taikobot robot, the variation in the limb center-of-mass position is linearized, and the spring model output is converted into limb length and orientation, which are then mapped to specific joint commands. Findings Simulations of two typical scenarios, point contact and line contact, were conducted on the PyBullet platform. The results demonstrate that the proposed method can achieve stable point-to-point locomotion while effectively suppressing undesired body rotation and detachment, validating its effectiveness and robustness. Originality/value This work introduces a novel adaptive mass-spring-based planning framework tailored to zero-gravity cabin environments, providing a computationally efficient way to map task-level maneuvering requirements to joint-level commands without solving complex inverse dynamics or encountering multiple-solution issues. It offers a practical motion planning approach for wall-reaction locomotion of space station robots.
Zhao et al. (Fri,) studied this question.