Imagine-then-Plan：基于世界模型自适应前瞻的智能体学习
原标题：Imagine-then-Plan: Agent Learning from Adaptive Lookahead with World Models
技术动态AI 65
来源：arXiv cs.AI发布时间待核实
arXiv:2601.08955v3 Announce Type: replace-cross Abstract: Recent advances in world models have shown promise for modeling future dynamics of environmental states, enabling agents to reason and act without accessing real environments. Current methods mainly perform single-step or fixed-horizon rollouts, leaving their potential for complex task planning under-exploited. We propose Imagine-then-Plan (\texttt{ITP}), a unified framework for agent learning via lookahead imagination, where an agent's p…