# Learning Quadruped Walking from Seconds of Demonstration

arXiv: https://arxiv.org/abs/2603.06961 · HTML: https://mlsanity.com/paper/2603.06961/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Real-world robot offline RL, one of 13 directions in Offline RL and imitation learning from data.
- Builds on 12 of its 21 references that are in the map.
- First posted on arXiv 7 Mar 2026.

## Placed in

- Offline RL and imitation learning from data / [Real-world robot offline RL](https://mlsanity.com/offlinerl/real-world-robot-offline-rl/)

## Builds on (most cited first, 12 of 12)

- [Learning Fine-Grained Bimanual Manipulation with Low-Cost Hardware](https://mlsanity.com/paper/2304.13705/) (2023)
- [Learning to Walk in Minutes Using Massively Parallel Deep Reinforcement Learning](https://mlsanity.com/paper/2109.11978/) (2021)
- [Learning Agile Robotic Locomotion Skills by Imitating Animals](https://mlsanity.com/paper/2004.00784/) (2020)
- [Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes](https://mlsanity.com/paper/2408.03539/) (2024)
- [Highly Dynamic Quadruped Locomotion via Whole-Body Impulse Control and Model Predictive Control](https://mlsanity.com/paper/1909.06586/) (2019)
- [DiffuseLoco: Real-Time Legged Locomotion Control with Diffusion from Offline Datasets](https://mlsanity.com/paper/2404.19264/) (2024)
- [Towards Learning to Imitate from a Single Video Demonstration](https://mlsanity.com/paper/1901.07186/) (2019)
- [SDS -- See it, Do it, Sorted: Quadruped Skill Synthesis from Single Video Demonstration](https://mlsanity.com/paper/2410.11571/) (2024)
- [From Bipedal Walking to Quadrupedal Locomotion: Full-Body Dynamics Decomposition for Rapid Gait Generation](https://mlsanity.com/paper/1909.08560/) (2019)
- [Activation-Descent Regularization for Input Optimization of ReLU Networks](https://mlsanity.com/paper/2406.00494/) (2024)
- [GRoQ-LoCO: Generalist and Robot-agnostic Quadruped Locomotion Control using Offline Datasets](https://mlsanity.com/paper/2505.10973/) (2025)
- [Simultaneous Contact Sequence and Patch Planning for Dynamic Locomotion](https://mlsanity.com/paper/2508.12928/) (2025)

## Abstract (arXiv)

Quadruped locomotion provides a natural setting for understanding when model-free learning can outperform model-based control design, by exploiting data patterns to bypass the difficulty of optimizing over discrete contacts and the combinatorial explosion of mode changes. We give a principled analysis of why imitation learning with quadrupeds can be inherently effective in a small data regime, based on the structure of its limit cycles, Poincar'e return maps, and local numerical properties of neural networks. The understanding motivates a new imitation learning method that regulates the alignment between variations in a latent space and those over the output actions. Hardware experiments confirm that a few seconds of demonstration is sufficient to train various locomotion policies from scratch entirely offline with reasonable robustness.