arXiv cs.AI· Royson Lee, Fady Rezk, Titouan Parcollet, Timothy Hospedales, Cristina Cornelio·· 3 小时前AI 评分40
Metro-WM：基于可实现子目标的长时程潜空间规划
Metro-WM: Long-Horizon Latent Planning with Realisable Sub-Goals
AI 导读
Metro-WM 是一种分层规划框架，针对 JEPA 模型预测控制中潜空间子目标常不可实现的缺陷，改为从历史经验中检索真实状态作为子目标。它以离线专家演示或随机动作轨迹的观测帧构建图并进行跨轨迹拼接规划，执行偏差时可即时重寻最优路径。实验显示其长时程成功率最高超出最强分层基线 37.33 个百分点，速度快 10.9 倍，离线计算量少 13-56 倍。
正文 · 原文
Abstract:Model-predictive control with Joint-Embedding Predictive Architectures (JEPAs) provides a strong zero-shot goal-reaching planner, but it is only effective over short planning horizons. Hierarchical extensions attempt to bridge this gap by learning a macro planner to predict intermediate latent sub-goals to guide the micro planner. In this work, we demonstrate that unconstrained latent sub-goal prediction is fundamentally flawed. A rigorous evaluation reveals that leading hierarchical macro planners routinely emit physically unrealisable sub-goals. To resolve this, we introduce Metro-WM, a hierarchical framework that issues sub-goals by retrieving genuine states from prior experience rather than generating ungrounded latent vectors. Specifically, Metro-WM constructs a graph whose vertices are observed frames from offline expert demonstrations or random-action trajectories, allowing frames from different episodes to be connected and stitched into routes to the goal. Planning over the full graph also makes the system highly robust to execution errors: if the micro planner drifts off course, Metro-WM instantly finds a new optimal path from the current state. Our experiments show that Metro-WM achieves superior long-horizon success rates of up to 37.33 percentage points over the strongest hierarchical baseline on the same world model while being up to 10.9 times faster, requiring both 13-56 times less offline compute and fewer tuned hyperparameters. Additional analysis reveals that Metro-WM frequently finds shorter paths than the offline demonstrations, outperforms an oracle relying on the query's own demonstration on most benchmarks, and maintains robust performance under extremely sparse dataset conditions.
| Subjects: | Artificial Intelligence (cs.AI); Robotics (cs.RO) |
| Cite as: | arXiv:2609.31868 [cs.AI] |
| (or arXiv:2609.31868v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2609.31868 arXiv-issued DOI via DataCite |
Submission history
From: Royson Lee [view email]
[v1]
Fri, 25 Sep 2026 18:09:49 UTC (1,263 KB)
[v2]
Fri, 16 Oct 2026 15:23:50 UTC (1,263 KB)
[v3]
Fri, 23 Oct 2026 15:10:39 UTC (2,033 KB)
来源：arXiv cs.AI · arxiv.org