发现世界模型长程预测失败源于运动学而非动力学错误
Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure
- 提出运动学一致性诊断，量化想象轨迹偏离物理规律程度
- 实测显示模型想象误差比真实物理轨迹高两个数量级
- 适合研究长时序生成与世界模型可靠性的研究人员
长时序世界模型的失败常被归因于误差累积，但这种说法未区分误差类型。本文提出运动学与动力学的重审视角：世界模型更倾向于想象运动学而非动力学行为。为此，我们定义了想象运动学一致性误差（iKCE），作为每步的诊断指标，衡量滚动预测偏离闭式运动学解的程度，并设计扰动实验测试当物理条件跨过相变边界时iKCE是否响应。在公开发布的DreamerV3模型（训练于DMC walker-walk数据集）上，想象iKCE值约为真实物理滚动的两倍量级。在摩擦系数扫描跨越步态崩溃边界时，模型的奖励显著下降，但其iKCE保持统计平稳，展现出典型的运动学而非动力学特征。该诊断可有效区分超过身体步态周期长度的长时序预测中的运动学与动力学想象。
原文摘要 · Abstract (English)
Long-horizon failure in world models is conventionally attributed to compounding error, a generic framing that does not distinguish what kind of error compounds. We propose a kinematic-vs-dynamic reframing: world models tend to imagine kinematically rather than dynamically. We operationalize this as the imagined Kinematic-Consistency Error, a per-step diagnostic that measures how far a rollout departs from a closed-form kinematic null, paired with a perturbation protocol that tests whether iKCE responds when physical conditions cross a regime boundary. We instantiate the diagnostic on a released DreamerV3 checkpoint trained on DMC walker-walk, where imagined iKCE runs roughly two orders of magnitude above that of matched real-physics rollouts. Across a friction sweep that crosses the gait-collapse boundary, the model's iKCE stays statistically flat even as the trained policy's reward collapses through the same range, providing the kinematic-not-dynamic signature. The diagnostic distinguishes kinematic from dynamic imagination at horizons longer than the embodiment's gait period.
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。