让视觉预测在多步行动中保持一致，提升机器人导航的可靠性。
MWM: Mobile World Models for Action-Conditioned Consistent Prediction
- 分两阶段训练：先结构预训练，再动作一致性微调。
- 少步扩散蒸馏后，多步预测误差降低37%以上。
- 适合需要稳定视觉预测的机器人规划任务。
世界模型可实现基于想象未来状态的规划，为具身导航提供有效框架。然而，现有导航世界模型常缺乏动作条件一致性，导致多步推演时视觉预测仍会漂移，影响规划效果。同时，高效部署需少步扩散推理，但现有蒸馏方法未显式保留推演一致性，造成训练与推理不匹配。为此，我们提出MWM，一种面向基于规划的图像目标导航的轻量级世界模型。通过两阶段训练框架：先结构预训练，再引入动作条件一致性（ACC）后训练，显著提升动作条件下的推演一致性。进一步提出推理一致性状态蒸馏（ICSD），实现少步扩散蒸馏并增强推演一致性。在基准和真实世界任务上的实验表明，该方法在视觉保真度、轨迹准确率、规划成功率及推理效率上均取得持续提升。代码与主页见：https://github.com/AIGeeksGroup/MWM，https://aigeeksgroup.github.io/MWM。
原文摘要 · Abstract (English)
World models enable planning in imagined future predicted space, offering a promising framework for embodied navigation. However, existing navigation world models often lack action-conditioned consistency, so visually plausible predictions can still drift under multi-step rollout and degrade planning. Moreover, efficient deployment requires few-step diffusion inference, but existing distillation methods do not explicitly preserve rollout consistency, creating a training-inference mismatch. To address these challenges, we propose MWM, a mobile world model for planning-based image-goal navigation. Specifically, we introduce a two-stage training framework that combines structure pretraining with Action-Conditioned Consistency (ACC) post-training to improve action-conditioned rollout consistency. We further introduce Inference-Consistent State Distillation (ICSD) for few-step diffusion distillation with improved rollout consistency. Our experiments on benchmark and real-world tasks demonstrate consistent gains in visual fidelity, trajectory accuracy, planning success, and inference efficiency. Code: https://github.com/AIGeeksGroup/MWM. Website: https://aigeeksgroup.github.io/MWM.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。