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

TL;DR

This paper introduces a data-driven framework for humanoid robot design that leverages existing mechanical designs and human motion data to create a compact, geometry-preserving latent space for efficient optimization.

Contribution

It proposes a novel paradigm that minimizes human involvement by learning design spaces from existing data and defining loss functions from motion data, enabling automated robot design.

Findings

Constructed a geometry-preserving latent space for humanoid upper body designs.

Demonstrated effective optimization of robot designs using gradient-free methods.

Leveraged existing designs and human motion data to guide automated discovery.

Abstract

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: (i) the vast, unstructured design space and (ii) the difficulty of constructing task-specific loss functions. We propose a new paradigm that minimizes human involvement by (i) learning the design search space from existing mechanical designs, rather than hand-crafting it, and (ii) 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.