Search papers, labs, and topics across Lattice.
Humanoid-GPT is a novel GPT-style Transformer designed for whole-body motion tracking, trained on a massive 2 billion-frame corpus that integrates various motion capture datasets. This approach overcomes limitations of previous models by achieving zero-shot generalization to unseen motions and control tasks, demonstrating enhanced performance in tracking dynamic behaviors. Extensive experiments reveal that Humanoid-GPT sets a new benchmark in motion tracking, showcasing its ability to handle complex tasks without prior exposure.
Humanoid-GPT achieves unprecedented zero-shot generalization in motion tracking, outperforming traditional models by leveraging a billion-scale motion dataset.
We introduce Humanoid-GPT, a GPT-style Transformer with causal attention trained on a billion-scale motion corpus for whole-body control. Unlike prior shallow MLP trackers constrained by scarce data and an agility-generalization trade-off, Humanoid-GPT is pre-trained on a 2B-frame retargeted corpus that unifies all major mocap datasets with large-scale in-house recordings. Scaling both data and model capacity yields a single generative Transformer that tracks highly dynamic behaviors while achieving unprecedented zero-shot generalization to unseen motions and control tasks. Extensive experiments and scaling analyses show that our model establishes a new performance frontier, demonstrating robust zero-shot generalization to unseen tasks while simultaneously tracking highly dynamic and complex motions.
Tsinghua AI