arXiv cs.AI· Yuhai Wang, Jiawei Xia, Bowen Sun, Rongxuan Zhou, Xiao Hu, Yongliang Shi, Jing Du, Yang Ye·· 4 小时前AI 评分37
PRISM：利用世界模型先验引导的想象力采样进行目标条件规划
PRISM: PRior-guided Imagination Sampling in world Models
AI 导读
PRISM 研究如何从冻结的 latent world model 表征中提取动作直觉并用其引导目标条件规划：一个轻量级动作头直接从冻结编码器特征学习目标条件动作先验，其均值与预测不确定性引导动作搜索，由世界模型评估动作后果。在四个仿真控制任务和两个真实机器人平台上的实验表明，该动作头提升了规划性能，且同一融合接口无需修改即可接受扩散先验。
正文 · 原文
Abstract:Latent world models learn representations that enable them to predict how actions change the environment - a capability we call physical intuition. We find that these same representations also support action intuition: a lightweight head can learn a goal-conditioned action prior directly from the frozen encoder's features. We introduce PRISM to investigate how action intuition can be extracted from frozen world-model representations and used to guide goal-conditioned planning. The prior's mean and predicted uncertainty guide the search for useful actions, while the world model evaluates their consequences against the goal. Experiments on four simulated control tasks and two real-robot platforms show that a lightweight action head improves planning performance, and that the same fusion interface accepts a diffusion prior over the same frozen features without modification. These results demonstrate the potential of reusing world-model representations to both predict action consequences and guide action selection.
| Subjects: | Robotics (cs.RO); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2606.07974 [cs.RO] |
| (or arXiv:2606.07974v2 [cs.RO] for this version) | |
| https://doi.org/10.48550/arXiv.2606.07974 arXiv-issued DOI via DataCite |
Submission history
From: Yang Ye [view email]
[v1]
Sat, 6 Jun 2026 04:18:10 UTC (11,128 KB)
[v2]
Sat, 24 Oct 2026 23:45:30 UTC (4,118 KB)
来源：arXiv cs.AI · arxiv.org