Open-HumanScale, 5만 시간 인간 영상으로 로봇 기초 모델 학습
Open-HumanScale: An Open Recipe for Scaling Human and Robot Experience for Embodied Foundation Models
- 매체
- Hugging Face
- 게시
- 중요도
- 4/5
- 화제 지수
- 11
한국어 요약
인간의 1인칭 조작 영상에서 추출한 학습 신호로 로봇 정책을 확장하는 Open-HumanScale을 제안했다. 손 동작과 자연어로 주석을 단 Open-HumanScale-50K 말뭉치 5만 시간으로 전학습한 HumanWAM-Vanilla와 HumanWAM-Goal은 로봇 미세조정 후 시뮬레이션과 실제 로봇 벤치마크에서 92% 이상의 성공률을 기록했다. 저자들은 데이터 큐레이션·전학습 파이프라인과 모델 가중치, 추론 코드를 공개 배포했다.
같은 소식, 다른 매체 · 2곳
원문 발췌 · English
Egocentric human video offers a scalable source of manipulation priors for robot learning, yet translating raw data scale into generalizable robot policies demands rigorous data curation and training recipes. We introduce Open-HumanScale, the first fully open recipe for scaling egocentric pretraining to 50,000 hours, anchored by Open-HumanScale-50K---a curated egocentric video corpus densely annotated with hand motion and natural language. Leveraging this corpus, we systematically investigate pure human pretraining via closed-loop simulation evaluations and uncover an empirical log-linear scaling law linking pretraining data hours to downstream robot success rates. Grounded in these findings, we develop HumanWAM-Vanilla, a world--action model pretrained exclusively on 50k hours of egocentric human video. Upon robot post-training, HumanWAM-Vanilla reaches 92.08% success on RoboTwin 2.0 Clean2Clean and 68.88% on Clean2Random, alongside a 14.17% mean success rate and a 19.00 mean progress score on RoboDojo. To push the capability frontier further, we explore joint human--robot pretraining with HumanWAM-Goal, a visual-goal-conditioned world--action model. Hybrid data scaling unlocks broad zero-shot generalization across both simulation and physical benchmarks, propelling post-trained RoboTwin 2.0 performance to new state-of-the-art levels: 92.70% on Clean2Clean and a substantial leap to 84.60% on Clean2Random. On physical robots, HumanWAM-Goal further exhibits emergent embodied in-context learning, dynamically adapting its manipulation behavior to unseen visual goals without parameter updates. To support reproducible scaling research in embodied AI, we open-source the Open-HumanScale-50K corpus, data curation and pretraining pipelines, alongside the pretrained and post-trained HumanWAM checkpoints, inference code, and simulation evaluation suites.