World Models That Know When They Don't Know: Controllable Video Generation with Calibrated Uncertainty

2025-12-05

AI Summary

Controllable video generation models often produce physically implausible frames—“hallucinations”—severely undermining their reliability for robot policy evaluation and planning. Existing approaches lack calibrated uncertainty estimation, hindering identification of untrustworthy generated regions. To address this, we propose C3, the first framework enabling sub-block-level, continuous-valued hypothetical uncertainty quantification: it models uncertainty in latent space to ensure training stability and maps it to pixel space to generate high-resolution uncertainty heatmaps; correctness and calibration are jointly optimized using strict proper scoring rules. Evaluated on Bridge and DROID datasets, C3 achieves not only in-distribution calibration but also substantially improved out-of-distribution detection performance. C3 thus provides the first fine-grained, trustworthy, dense confidence estimation scheme for large-scale video generation models.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Computer Vision: Large Vision Models

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent advances in generative video models have led to significant breakthroughs in high-fidelity video synthesis, specifically in controllable video generation where the generated video is conditioned on text and action inputs, e.g., in instruction-guided video editing and world modeling in robotics. Despite these exceptional capabilities, controllable video models often hallucinate - generating future video frames that are misaligned with physical reality - which raises serious concerns in many tasks such as robot policy evaluation and planning. However, state-of-the-art video models lack the ability to assess and express their confidence, impeding hallucination mitigation. To rigorously address this challenge, we propose C3, an uncertainty quantification (UQ) method for training continuous-scale calibrated controllable video models for dense confidence estimation at the subpatch level, precisely localizing the uncertainty in each generated video frame. Our UQ method introduces three core innovations to empower video models to estimate their uncertainty. First, our method develops a novel framework that trains video models for correctness and calibration via strictly proper scoring rules. Second, we estimate the video model's uncertainty in latent space, avoiding training instability and prohibitive training costs associated with pixel-space approaches. Third, we map the dense latent-space uncertainty to interpretable pixel-level uncertainty in the RGB space for intuitive visualization, providing high-resolution uncertainty heatmaps that identify untrustworthy regions. Through extensive experiments on large-scale robot learning datasets (Bridge and DROID) and real-world evaluations, we demonstrate that our method not only provides calibrated uncertainty estimates within the training distribution, but also enables effective out-of-distribution detection.

Problem

• Addresses hallucination in controllable video generation models
• Introduces uncertainty quantification for dense confidence estimation
• Enables calibrated uncertainty in subpatch-level video frame generation

Innovation

• Calibrated uncertainty quantification for video generation
• Latent-space uncertainty estimation avoiding training instability
• Pixel-level uncertainty mapping for intuitive visualization