arXiv：cs.LG· Ziang Fu, Ning Ning·· 4 小时前AI 评分35
Control-Geometry Straightening：让基于采样的潜在规划更高效
Control-Geometry Straightening for Sampling-Based Latent Planning
AI 导读
研究人员提出 Control-Geometry Straightening（CGS），一种通过直接拉直控制几何来学习规划器友好表征的辅助损失，仅用像素-动作对的局部转移即可对齐动作间与潜在差异间的成对余弦相似度。
正文
Abstract:Joint-embedding predictive architectures enable planning with latent world models, but accurate transition prediction alone does not ensure that the planning objective is easy to optimize. We introduce Control-Geometry Straightening (CGS), a single auxiliary loss that learns planner-friendly representations by directly straightening control geometry for sampling-efficient planning. CGS matches pairwise cosine similarities among actions to those among corresponding latent differences only using local transitions from pixel-action pairs. The loss can be applied across world-model architectures using end-to-end learned or pretrained representations. Under linear-dynamics, our theoretical analysis connects this objective to temporal straightening and more balanced terminal-cost curvature across the full planning horizon, yielding finite-budget guarantees for MPPI, local contraction results for CEM, and convergence bounds for gradient descent. Across four control environments and multiple planners, CGS improves planning with fewer sampled candidates and refinement steps, achieving success-rate gains up to 20 and 12.6 percentage points over LeWorldModel (LeWM) and its temporal-straightening variant (LeWM+TS), respectively, with sampling-based planners using 128 candidates per update. Probes, comparisons with DINO-WM architecture, and planner-side ablations clarify how latent motion organization, state dependence, and dynamical context shape planning behavior. Straightening control geometry thus makes good action sequences easier to find under limited planning budgets.
| Subjects: | Machine Learning (cs.LG); Machine Learning (stat.ML) |
| Cite as: | arXiv:2609.35603 [cs.LG] |
| (or arXiv:2609.35603v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2609.35603 arXiv-issued DOI via DataCite |
Submission history
From: Ziang Fu [view email]
[v1]
Mon, 28 Sep 2026 16:51:33 UTC (2,486 KB)
[v2]
Wed, 7 Oct 2026 15:09:30 UTC (2,486 KB)
来源：arXiv：cs.LG · arxiv.org