ALERT: Extending Humanoid Motion Trackers with a General and Lightweight Human Avoidance Pipeline
Organizations: Institute of Automation, Chinese Academy of Sciences · University of Chinese Academy of Sciences · Beijing Institute of Technology
Abstract
Whole-body motion tracking has enabled humanoid dancing, athletic motions, and teleoperation, providing a foundation for capability demonstrations and manipulation data collection. However, existing tracking policies lack awareness of nearby humans and proactive avoidance capabilities, while motion substitution and local constraint-based corrections may compromise reference motion integrity and tracking continuity. We propose ALERT, a general lightweight pipeline for extending motion tracking to shared human-robot environments. Through lightweight adaptation, it integrates human perception and proactive avoidance into pretrained tracking policies, balancing avoidance demands with reference motion tracking while preserving existing motion capabilities. Simulation and real-world experiments across tracking backbones demonstrate reliable collision-free execution through localized human avoidance, preserving whole-body coordination and enabling rapid recovery of reference tracking, while retaining the ability of pretrained policies to track unseen motions.
Figures & tables
| Method | |||||||
|---|---|---|---|---|---|---|---|
| SONIC | |||||||
| Original | |||||||
| Ours (w/o H) | / | ||||||
| Ours (w/ H) | |||||||
| TWIST2 | |||||||
| Original | |||||||
| Motion | |||
|---|---|---|---|
| Ninja Dance | |||
| Disco Dance | |||
| Kick | |||
| Teleop |
| Method | SR | |||
|---|---|---|---|---|
| SONIC | 37.29_{{\color[rgb]{0.5,0.5,0.5}\pm 0.42}} | 35.87_{{\color[rgb]{0.5,0.5,0.5}\pm 1.79}} | 0.00_{{\color[rgb]{0.5,0.5,0.5}\pm 0.00}} | |
| CM | 142.08_{{\color[rgb]{0.5,0.5,0.5}\pm 20.03}} | 51.76_{{\color[rgb]{0.5,0.5,0.5}\pm 12.07}} | 1.27_{{\color[rgb]{0.5,0.5,0.5}\pm 0.38}} | |
| CSF | 172.18_{{\color[rgb]{0.5,0.5,0.5}\pm 28.89}} | 38.26_{{\color[rgb]{0.5,0.5,0.5}\pm 10.21}} | 2.19_{{\color[rgb]{0.5,0.5,0.5}\pm 0.45}} | |
| w/o LoRA | 110.84_{{\color[rgb]{0.5,0.5,0.5}\pm 15.83}} | 68.64_{{\color[rgb]{0.5,0.5,0.5}\pm 8.39}} | 2.04_{{\color[rgb]{0.5,0.5,0.5}\pm 2.56}} | |
| w/o PF | 112.30_{{\color[rgb]{0.5,0.5,0.5}\pm 14.64}} | 44.39_{{\color[rgb]{0.5,0.5,0.5}\pm 15.01}} | 1.83_{{\color[rgb]{0.5,0.5,0.5}\pm 2.17}} | |
| w/o Flow | 89.66_{{\color[rgb]{0.5,0.5,0.5}\pm 20.55}} | 50.66_{{\color[rgb]{0.5,0.5,0.5}\pm 16.36}} | 1.99_{{\color[rgb]{0.5,0.5,0.5}\pm 1.48}} |
Humanoid Motion Tracking
Humanoid motion tracking trains robots to reproduce human-like whole-body movements while remaining physically stable and feasible.
See all 91 papers