SPRINT: Efficient Spectral Priors for Humanoid Athletic Sprints

2026-05-27

AI Summary

This work addresses the challenges of high-speed sprinting in humanoid robots, which are hindered by the lack of feasible motion references and stability difficulties at high velocities. The authors propose the SPRINT framework, which introduces a frequency-adaptive spectral prior that models the periodicity of human gait to generate natural joint trajectories across a wide speed range from only sparse discrete motion sequences. This prior guides a reinforcement learning policy to achieve stable running. The approach enables zero-shot sim-to-real transfer and extrapolation beyond reference speeds, achieving a peak velocity of 6 m/s on the Unitree G1 platform. Furthermore, it supports smooth gait transitions and demonstrates successful zero-shot deployment in unstructured outdoor environments.

Abstract

The pursuit of humanoid athletic sprints is hindered by a scarcity of humanoid-viable kinematic reference data and the inability of existing frameworks to maintain stability during sprints. To overcome these limitations, we introduce SPRINT, a novel framework driven by efficient, frequency-adaptive spectral priors. By characterizing the fundamental periodicity of human locomotion in the frequency domain using a reference library of five discrete motion sequences, these priors generate kinematically feasible joint trajectories across a broad velocity spectrum, successfully extrapolating to speeds that exceed the reference distribution. Guided by these pretrained priors, the SPRINT policy achieves zero-shot sim-to-real transfer in field experiments on the Unitree G1 platform, reaching a peak sprinting velocity of 6 m/s and demonstrating seamless gait transitions while preserving biomimetic naturalness. Ultimately, this work establishes frequency-adaptive spectral priors as a highly data-efficient foundation for humanoid athletic sprints. The project page is available at https://anonymous.4open.science/w/SPRINT-138A/.

Problem

• humanoid sprinting
• kinematic reference data
• stability
• locomotion
• sim-to-real transfer

Innovation

• spectral priors
• frequency-adaptive
• humanoid sprinting
• zero-shot sim-to-real
• kinematic trajectory generation