Learning World Models from Egocentric Video for Home Robots
暂无中文版，显示英文原文。
30 秒速读
A video world model trained on robot egocentric footage predicts outcomes of actions for evaluation and planning.
研究问题
Can world models replace physical evaluation?
问题
Evaluating home robots is slow and unsafe.
既有方法
Physical rollouts or hand-built simulators.
新方法
Action-conditioned video prediction.
结果
Correlates with real success rates on household tasks.
局限
Hallucinated physics; single robot.
产业影响
Positions video generation vendors as robotics infrastructure.