With the growing volume and complexity of crewed space missions, astronauts face heavy workloads and safety risks when performing inspection, logistics and experimental operations inside confined space-station cabins. To address this, we propose a spherical modular self-reconfigurable free-flying robot specifically designed for micro-gravity cabins. Guided by TRIZ theory, a compact mechanical architecture is developed that reconciles the contradictions among thrust, docking precision and volume constraints. The robot employs six orthogonally arranged ducted-fan thrusters for full six-degree-of-freedom maneuvering and a hybrid mechanical–electromagnetic docking mechanism that enables reliable multi-robot assembly within large pose and position tolerances. A unified Lagrangian model is then established for both the free-flying platform and the post-reconfiguration multi-body system with an attached manipulator. To cope with nonlinear, strongly coupled dynamics and parameter uncertainties, a T–S fuzzy neural-network PID controller is designed to realize on-line tuning of PID gains. Comparative simulations with classical PID and fuzzy-PID schemes show that the proposed controller significantly shortens settling time, reduces overshoot and improves steady-state accuracy, verifying the effectiveness and robustness of the overall structural design and control framework for intra-vehicular service robots.
Tan et al. (2026) studied this question.