LEGO: Latent-space Exploration for Geometry-aware Optimization of Humanoid Kinematic Design

Designing robot morphologies and kinematics has traditionally relied on human intuition, with little systematic foundation. Motion-design co-optimization offers a promising path toward automation, but two major challenges remain: the vast, unstructured design space and the difficulty of constructing task-specific loss functions. We propose a new paradigm that minimizes human involvement by learning the design search space from existing mechanical designs, rather than hand-crafting it, and defining the loss directly from human motion data via motion retargeting and Procrustes analysis.

Using screw-theory-based joint axis representation and isometric manifold learning, we construct a compact, geometry-preserving latent space of humanoid upper body designs in which optimization is tractable. We then solve design optimization in this latent space using gradient-free optimization. Our approach establishes a principled framework for data-driven robot design and demonstrates that leveraging existing designs and human motion can effectively guide the automated discovery of novel robot design.

Our work has been accepted in ICRA 2026 and can be cited as arXiv:2604.08636 [cs.RO] or arXiv:2604.08636v1 [cs.RO] for this version, with a DOI of https://doi.org/10.48550/arXiv.2604.08636, which was issued by arXiv via DataCite, pending registration. The submission history of our work is available, with the first version submitted by Jihwan Yoon on Thu, 9 Apr 2026 17:06:50 UTC, with a file size of 52,481 KB.