INTACT：用于无搜索世界模型的同构意图到行为学习

摘要

前向潜在世界模型预测动作如何改变场景，但仅通过昂贵的测试时搜索才能为期望变化恢复动作。我们提出 INTACT（INtent-To-ACTion，意图到动作），一种端到端 JEPA，将带动作标签、无奖励的轨迹转化为可部署的意图到动作接口。每次转移提供物理意图 z_{t+1}−z_t，而未来目标提供部署意图 sg(z_g)−z_t。该架构在局部与目标运动意图主干-输入图之间通过相同的四槽语法和共享参数保持同构，并通过由同一预测器诱导的动作法则语义（而非逐点潜在等价）在受支持的局部与目标运动意图族之间保持同构。INTACT 还提供从 RGB 证据到动作有效的潜在意图坐标、以及从意图族到其对应动作法则族的完整迁移。非对称端点梯度锚定物理后继并将未来目标固定为锚点，在不进行逐点潜在匹配或全局线性动力学的情况下联结表征学习与控制。所得坐标支持稳健的分布式动作法则：其条件均值可直接作为免搜索策略，同时采样仍可用于多样性或可选验证。在 LeWM 的四个官方任务上，单轮训练、零搜索模型达到 85.78%、100.00%、97.67% 和 97.89% 的成功率。以 Direct 计划为中心的可选局部 CEM 使用 384 个而非 9,000 个候选序列达到 96.86% 的宏观成功率，采样量减少 23.44 倍，同时较纯 CEM 提升 16.00 个点。一个共享的四任务编码器达到 89.39% 的 E5 Direct 宏观平均，并在每个任务上均优于联合训练的 LeWM；预测-专家动作族 kNN 以 r=0.954 跟踪 Direct 成功率。Direct 推理耗时 2.9–5.5 ms。

English

Forward latent world models predict how actions change a scene, but recover actions for a desired change only through expensive test-time search. We introduce INTACT (INtent-To-ACTion), an end-to-end JEPA that turns action-labeled, reward-free trajectories into a deployable intent-to-action interface. Each transition supplies physical intent z_{t+1}-z_t, while a future goal supplies deployment intent sg(z_g)-z_t. The architecture is isomorphic between the local and goal motion-intent backbone-input graphs through an identical four-slot grammar and shared parameters, and between supported local and goal motion-intent families through action-law semantics induced by the same predictor rather than pointwise latent equality. INTACT also provides intact transfer from RGB evidence to action-effective latent intent coordinates and from intent families to their corresponding action-law families. Asymmetric endpoint gradients ground physical successors and fix future goals as anchors, joining representation learning and control without pointwise latent matching or globally linear dynamics. The resulting coordinates support a robust distributional action law: its conditional mean serves directly as a search-free policy, while sampling remains available for diversity or optional verification. On the four official LeWM tasks, one-epoch, zero-search models reach 85.78\%, 100.00\%, 97.67\%, and 97.89\% success. Optional local CEM centered on the Direct plan reaches 96.86\% macro success using 384 instead of 9,000 candidate sequences, reducing sampling by 23.44times while improving pure CEM by 16.00 points. One shared four-task encoder reaches 89.39\% E5 Direct macro and improves every task over jointly trained LeWM, while predicted--expert action-family kNN tracks Direct success at r=0.954. Direct inference takes 2.9--5.5 ms.