In a microgravity environment, modular self-reconfigurable robots can perform a range of on-orbit missions including solar-array deployment, serial-arm assembly, and failed-subsystem replacement, owing to their modular scalability and morphological versatility, tailored to mission-specific constraints and extended across these tasks. However, conventional cubic modules have rotational blind spots and pose-dependent interfaces that inflate alignment burden and trigger collisions and local deadlocks, especially for large-scale deployment. Due to the tight coupling of local motion feasibility in modular robotic systems, coupled with the connectivity and reachability requirements during reconfiguration, task allocation and decision sequencing for large-scale architecture are often NP-hard. To address these issues, we present an integrated reconfigurable hardware-algorithmic solution. Structurally, the concentric, nested spherical design with isotropic geometry and unified locking mechanism reduces sensitivity to pose alignment, mitigates collisions and deadlocks, and expands the reachable workspace. Algorithmically, reconfiguration planning is formulated as an integer programming problem, incorporating penalties to enforce connectivity and reachability constraints within a hierarchical framework. The top level determines the matching and reconfiguration sequence by the proposed Cross-correlation BFS-Tree Genetic Algorithm with Gaussian mutation (CBGA), and the lower level aims at path planning using the designed kinematics-aware parallel A*. Extensive simulation and experiments are conducted with varied number of modular robots. The results demonstrate that the proposed system maintains full connectivity and reachability while achieving rapid convergence with low relocation steps even for large-scale architecture. Such capability thereby establishes its practical viability for autonomous modular reconfiguration in on-orbit missions.
扫码关注我们
求助内容:
应助结果提醒方式:
