超越搜索的潜在几何：在世界模型中摊销规划
Latent Geometry Beyond Search: Amortizing Planning in World Models
AI总结：
提出在正则化潜在几何下，将规划摊销为潜在逆动力学映射，以轻量级GC-IDM替代在线搜索，在七个环境协议中匹配或超越CEM，决策成本降低100-130倍。
摘要：
现代基于视觉的世界模型可以将观测表示为紧凑而富有表现力的潜在流形，但在这些空间中进行快速的目标导向规划仍然具有挑战性。这引发了一个核心问题：学习到的表示何时简化控制，而不仅仅是实现预测？我们在预训练的LeWorldModel中研究这个问题，其潜在几何通过正则化实现平滑性和均匀性。我们的关键见解是，在这种几何下，规划可以摊销为潜在逆动力学映射，而无需在线搜索。因此，我们用一个轻量级的目标条件逆动力学模型（GC-IDM）替代迭代规划，该模型将当前潜在状态、目标潜在状态和剩余时间步直接映射到下一个动作。实验上，在涵盖导航、接触丰富的操作和连续控制的四个基准环境中，我们的控制器在八个环境-协议设置中的七个上匹配或超过了CEM，同时将每次决策成本降低了100-130倍。对测试时规划器（CEM、MPPI、iCEM和基于梯度的方法）的更广泛扫描表明，这一结果并非特定于某个优化器。这些发现表明，测试时规划恢复的大部分结构已经局部编码在潜在表示中。更广泛地说，我们的结果表明，足够结构化的潜在空间可以将部分规划负担从在线优化转移到学习推理。我们的代码公开在 https://github.com/hdnndh/Latent-Geometry-Beyond-Search-Amortizing-Planning-in-World-Models 。
英文摘要：
Modern vision-based world models can represent observations as compact yet expressive latent manifolds, but fast goal-oriented planning in these spaces remains challenging. This raises a central question: when does a learned representation simplify control, rather than merely enabling prediction? We study this question in a pretrained LeWorldModel, whose latent geometry is regularized for smoothness and uniformity. Our key insight is that, under such geometry, planning can be amortized into a latent inverse-dynamics mapping instead of requiring online search. We therefore replace iterative planning with a lightweight Goal-Conditioned Inverse Dynamics Model (GC-IDM) that maps the current latent state, goal latent state, and remaining horizon directly to the next action. Empirically, across four benchmark environments spanning navigation, contact-rich manipulation, and continuous control, our controller matches or exceeds CEM in seven of eight environment-protocol settings while reducing per-decision cost by 100-130x. A broader sweep over test-time planners (CEM, MPPI, iCEM, and gradient-based methods) shows that this result is not specific to a particular optimizer. These findings suggest that much of the structure recovered by test-time planning is already locally encoded in the latent representation. More broadly, our results indicate that sufficiently structured latent spaces can shift part of the planning burden from online optimization to learned inference. Our code is publicly available at https://github.com/hdnndh/Latent-Geometry-Beyond-Search-Amortizing-Planning-in-World-Models .
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
