Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure

Hugging Face Daily Papers Papers

Summary

This paper diagnoses long-horizon failures in world models, attributing them to kinematic rather than dynamic imagination. The authors introduce a metric (iKCE) and show that imagined rollouts fail to capture dynamic regime changes even as policy rewards collapse.

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.
Original Article
View Cached Full Text

Cached at: 07/09/26, 03:40 PM

Paper page - Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure

Source: https://huggingface.co/papers/2607.05966

Abstract

World models exhibit long-horizon failures due to kinematic rather than dynamic imagination, as demonstrated by measuring imagined kinematic-consistency error which remains flat while policy rewards collapse across friction boundaries.

Long-horizon failure inworld modelsis conventionally attributed tocompounding error, a generic framing that does not distinguish what kind of error compounds. We propose akinematic-vs-dynamic reframing:world modelstend to imagine kinematically rather than dynamically. We operationalize this as the imaginedKinematic-Consistency Error, a per-step diagnostic that measures how far arolloutdeparts from aclosed-form kinematic null, paired with aperturbation protocolthat tests whetheriKCEresponds when physical conditions cross a regime boundary. We instantiate the diagnostic on a releasedDreamerV3checkpoint trained onDMC walker-walk, where imaginediKCEruns roughly two orders of magnitude above that of matched real-physicsrollouts. Across a friction sweep that crosses thegait-collapse boundary, the model’siKCEstays 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 theembodiment’s gait period.

View arXiv pageView PDFGitHub1Add to collection

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2607.05966 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.05966 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.05966 in a Space README.md to link it from this page.

Collections including this paper1

Similar Articles

Understanding Rollout Error in Graph World Models

arXiv cs.AI

This paper analyzes long-horizon rollout error in Graph World Models (GWMs), proposing a unified framework with dynamic edges and introducing Error-Aware GWM that uses spectral regularization, rollout consistency, and critical-node weighting to prevent divergence.

The Replay Gap: Static Evaluation of Model Switching in LLM Agents Scores the Wrong World

arXiv cs.LG

This paper demonstrates that replay-based static evaluation of model switching in LLM agents is fundamentally flawed: when swapping models mid-trajectory, the environment and subsequent actions diverge dramatically from logged trajectories, invalidating most benchmark results. The authors propose branching rollouts as a more faithful evaluation method and release their harness and trajectories.