GE-Sim 2.0: A Roadmap Towards Comprehensive Closed-loop Video World Simulators for Robotic Manipulation

arXiv:2605.27491 · 2026-05-28 공개 · arXiv · PDF

video-generation robot-manipulation policy-learning action-conditioned h100 closed-loop-simulation rollout-acceleration proprioceptive-state

Abstract

We introduce GE-Sim 2.0 (Genie Envisioner World Simulator 2.0), a closed-loop video world simulator for robotic manipulation. Building on the action-conditioned video generation framework of Genie Envisioner, GE-Sim 2.0 is re-trained on thousands of hours of real-world robot data spanning teleoperation, contact-rich interaction, and on-robot policy deployment, substantially improving action-following fidelity and trajectory coverage. On top of this foundation, three new modules close the loop from video simulation to policy learning: a state expert that decodes proprioceptive state from video latents to support next-chunk prediction by downstream VLA policies; a world judge that scores generated rollouts against task instructions, yielding machine-verifiable success signals and rewards in place of manual inspection; and an acceleration framework that delivers a 25-frame rollout in 2.3 seconds on a single H100, with up to 4* frame skipping at inference for long-horizon evaluation. GE-Sim 2.0 tops the public WorldArena leaderboard at only 2B parameters, outperforming both dedicated robotic world models and closed-source general video generators, and policies trained against its rollouts and rewards translate into measurable real-world gains, establishing GE-Sim 2.0 as a practical platform for scalable evaluation and closed-loop learning of manipulation policies.

한국어 요약

📋 한 줄 요약

**[Robotic World Simulator / VLA Policy]** GE-Sim 2.0이 robotic manipulation의 closed-loop video world simulator — state expert·world judge·acceleration framework 3 모듈로 video sim→policy learning 폐쇄 — WorldArena 1위, 25-frame rollout 2.3초.

🎯 핵심 기여도

💡 핵심 아이디어

Robotic manipulation의 video world simulator는 (1) state 디코딩, (2) 자동 reward 채점, (3) 실시간 가속의 3 모듈로 video simulation→policy learning loop을 폐쇄해야 하며, 실로봇 데이터 기반 재학습은 action fidelity·trajectory coverage를 substantial 개선해 small 모델로도 SOTA·실세계 transfer를 달성한다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: Robotic manipulation의 closed-loop 학습을 video world model로 가능하게 함, manual inspection 자동화(world judge)로 scalable 평가, 작은 모델 크기로 SOTA 달성. **한계**: Manipulation 중심으로 다른 로봇 task(locomotion·flight) 일반화 추가 검증, world judge의 task instruction 해석 정확성 의존, long-horizon rollout의 drift는 후속.

🚀 실용적 활용