AI Daybook
AI Daybook
오늘의 지면
논문Physical AI

LAW, 언어·행동·세계 모델을 통합한 Physical AI 에이전트

LAW: Unifying Language, Action, and World Modeling for Agentic Physical Intelligence

매체
arXiv
게시
중요도
3/5
화제 지수
—

한국어 요약

연구팀이 언어 추론·행동 생성·세계 모델을 의미론적 부분 작업 진행도로 통합한 LAW 프레임워크를 제시했다. 혼합 전문가 트랜스포머로 다중 모드 컴퓨팅을 수행하며, RoboTwin 50개 작업에서 94.21% 평균 성공률을 달성했고 OpenWAM-α 백본 사용 시 95.56%에 도달한다. 실제 로봇 3개 작업에서는 65% 성공률을 기록했다.

원문 발췌 · English

World-action models jointly predict actions and future world states, but long-horizon manipulation further requires language reasoning to plan subtask objectives, assess execution progress, and guide action selection. We present LAW, a closed-loop VLM--WAM framework that unifies language reasoning, action generation, and world modeling through semantic subtask progress. Within the LAW agent, a hybrid-expert vision-language model plans the remaining subtasks, estimates the values of and fuses actions proposed by a world-action model, and asynchronously assesses execution status to support continued execution and subtask transitions. Within the world-action model, Mixture-of-Transformer Experts integrates modality-specific and shared computation for joint video--action modeling. Dual-stream hierarchical visual memory maintains temporal context for both models, while subtask-local value learning aligns action evaluation with semantic objectives and preserves supervision from successful prefixes of failed trajectories. LAW achieves 94.21% average success across 50 RoboTwin tasks. Replacing the WAM backbone with OpenWAM-$alpha$ further raises the average success rate to 95.56%, establishing a new open-source state of the art on RoboTwin. Despite having only 2B parameters, our VLM also outperforms variants using Embodied-R1.5 and GPT-5.5 High. On three long-horizon real-robot tasks, LAW achieves 65% average success, compared with 28.33% for $pi_{0.5}$ and 25% for Fast-WAM.

#vla#world-model#robotics
원문 기사 읽기
전체 보기 →