Towards Embodiment Scaling Laws in Robot Locomotion

Bo Ai, Liu Dai, Nico Bohlinger, Dichen Li, Tongzhou Mu, Zhanxin Wu, K. Fay, Henrik I Christensen, Jan Peters, Hao Su

Keywords: Cross-Embodiment Learning, Robot Locomotion, Embodiment Generation, Robotic Foundation Models, Reinforcement Learning

TL;DR: We generate a dataset of ~1,000 robot embodiments and train a single locomotion policy that generalizes to diverse unseen embodiments in both simulation and the real world via cross-embodiment training.

Abstract: Developing generalist agents that can operate across diverse tasks, environments, and robot embodiments is a grand challenge in robotics and artificial intelligence. In this work, we focus on the axis of embodiment and investigate embodiment scaling laws—the hypothesis that increasing the number of training embodiments improves generalization to unseen ones. Using robot locomotion as a test bed, we procedurally generate a dataset of $\sim$1,000 varied embodiments, spanning humanoids, quadrupeds, and hexapods, and train generalist policies capable of handling diverse observation and action spaces on random subsets. We find that increasing the number of training embodiments improves generalization to unseen ones, and scaling embodiments is more effective in enabling embodiment-level generalization than scaling data on small, fixed sets of embodiments. Notably, our best policy, trained on the full dataset, zero-shot transfers to novel embodiments in the real world, such as Unitree Go2 and H1. These results represent a step toward general embodied intelligence, with potential relevance to adaptive control for configurable robots, co-design of morphology and control, and beyond.

Submission Number: 1