HEFT: Heavy-Payload Full-size Humanoid Teleoperation with Privileged Motion Guidance and Windowed Payload Curriculum
2026-07-02 • Robotics
Robotics
AI summaryⓘ
The authors created HEFT, a system that helps a large humanoid robot copy human movements while carrying heavy objects. They solved problems related to noisy virtual reality tracking and the robot's difficulty balancing with heavy loads. HEFT uses special learning techniques to turn imperfect VR data into realistic motions and gradually trains the robot to handle heavier weights safely. They tested HEFT on a full-size robot called L7, which successfully performed movements like walking and squatting while carrying up to 24kg.
humanoid robotteleoperationmotion trackingvirtual reality trackingpayload handlingrobot balancemotion retargetingmachine learningrobot locomotion
Authors
Chenxin Liu, Qingzhou Lu, Guangxiao Yang, Xuanyang Shi, Chenghan Yang, Yanjiang Guo, Jianyu Chen
Abstract
General motion tracking and teleoperation offer a promising path to scalable humanoid skill acquisition, yet most existing frameworks are validated on compact platforms or without real payload interaction, leaving full-size humanoids with real payloads largely unexplored. Scaling to full-size humanoids introduces two compounding challenges: their larger inertia and tighter balance margins make tracking highly sensitive to noise, drift, and retargeting errors from commodity VR trackers, while their payload potential remains largely underutilized. We present HEFT, a heavy-payload full-size humanoid teleoperation framework that addresses both challenges. HEFT learns from deployable noisy VR references with physically plausible reconstructed references through Privileged Motion Guidance (PMG), and uses a Windowed Payload Curriculum (WPC) with expert-guided payload caps to acquire robust heavy-payload tracking. We deploy HEFT on L7, a 175cm, 65kg humanoid. The robot tracks motions including turns, forward/backward locomotion, and squats under payloads up to 24kg.