Dimensional Optimization of a Space Quadruped Climbing Robot Based on Dominant Joint-Workspace Partition Framework

Abstract This article presents a comprehensive dimensional design and optimization strategy for a redundantly actuated space quadruped climbing robot. Our approach focuses on non-dimensional optimization, utilizing a dimensionally normalized Jacobian and a design space dimensionality-reduction method for both the leg mechanisms and the overall robot structure. A key component of this strategy is a novel performance evaluation framework, the dominant joint-workspace partitioning framework (DJ-WPF). The DJ-WPF uses a joint participation index to identify dominant actuators for different regions of the workspace, enabling the selection of optimal region-specific actuator sets. Using this framework, we generate performance atlases to visualize key indicators and identify high-performing design regions. An optimal set of non-dimensional parameters is then determined using a weighted averaging method. Finally, these parameters are scaled to physical dimensions under practical constraints. The resulting optimized design achieves a condition index and a minimum payload index that are 50.19% and 79.3% of their theoretical maximums, respectively, while reducing computational cost.