Learning Humanoid Navigation From Human Data

Weizhuo Wang

doi:10.1109/lra.2026.3710346

Learning Humanoid Navigation From Human Data

2026-07-06

0

PRE

AI

DOI：10.1109/lra.2026.3710346

Original

Original request for help

Save

Abstract

En 中文

We present EgoNav, a system that enables a humanoid robot to perform zero-shot, goal-free traversal across evaluated unseen environments by learning entirely from 5 hours of human walking data, with no robot data or finetuning. A diffusion model predicts distributions of plausible future trajectories conditioned on past trajectory, a 360° visual memory fusing color, depth, and semantics, and video features from a frozen DINOv3 backbone that capture appearance cues invisible to depth sensors. A hybrid sampling scheme achieves real-time inference in 10 denoising steps, and a receding-horizon controller selects paths from the predicted distribution. We validate EgoNav through offline evaluations, real-world comparisons to VFH+ and MPPI, and failure-mode analysis during zero-shot deployment on a Unitree G1 humanoid across unseen indoor and outdoor environments. Behaviors such as waiting for doors to open, navigating around crowds, and avoiding glass walls emerge naturally from the learned prior. We will release the dataset and trained models.

Keywords:

Learning from demonstration

humanoid robot systems

vision-based navigation

Learning Humanoid Navigation From Human Data

Abstract

Journal

Organization