Driving · 2026-10-02
Completion Aware Guidance for World Action Models
Seungyeon Kim, Junhoo Lee, Baekseung Kim, Minkyu Kim, Nojun Kwak
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
World Action Models can generate visually plausible, action-consistent futures yet still omit the state transition needed to complete the task (“task-incomplete imagination”). The authors attribute this to an objective mismatch between fixed-horizon video generation and short-chunk control. They propose Completion Aware Guidance (CAG), a training-free sampling method that guides instruction conditioning toward completion, improving success and reducing task-incomplete imagination.
Why it matters
CAG provides an inference-time fix for a failure mode where WAM imagination stays locally plausible but fails to execute the completion transition required by the current task phase. The paper argues the backbone can be capable, and that the issue emerges from short-chunk control; CAG targets this at sampling time without retraining.
Method
- Identify task-incomplete imagination: WAM rollouts can be instruction-relevant and action-consistent while omitting the transition required for task completion.
- Introduce Completion Aware Guidance (CAG), a training-free sampling-time method that strengthens instruction conditioning relevant to the current prediction to elicit completion transitions while staying close to the WAM prior via a KL term.
- Use cross-attention pathways to compute token-wise scaling bounded by sparse gating, applying the intervention to bias the next video-action chunk toward phase completion.
Limitation
CAG is currently limited to sampling-time intervention, and extending completion aware guidance to WAM training is a promising direction for future work.
Abstract (from arXiv)
World Action Models (WAMs) predict visual futures and robot actions, yet they remain susceptible to task-incomplete imagination, where plausible, action-consistent predictions omit the transition needed for task completion. In this paper, we show that this failure is not inherent to the world model backbone, but emerges when adapted for short-chunk control, which can repeatedly favor plausible local continuations over task-completing transitions. To address this, we introduce Completion Aware Guidance (CAG), a training-free sampling method that guides generation toward task completion. Across representative WAMs, CAG improves success from 64% to 70% on a RoboTwin 2.0 subset and from 69% to 75% in zero-shot simulation, while reducing task-incomplete imagination from 79% to 40%.
Related papers
- FLEX-WAM: Flexible Block-Causal World-Action Models for Long-Horizon Imagination and Planning
- Achieve What You Imagined: Learning to Align Actions with Visual Plans
- The Planning Limits of Latent World Models
- Commit While Futures Agree: Consequence-Aware Adaptive Action Chunking for Robot Manipulation
- {\Delta}WAM: Distilling Action Tangent Fields into World Action Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-01 · Indexed by Pub-AI: 2026-10-02 UTC
Completion Aware Guidance for World Action Models
Preprint posted: 2026-10-01 · Original record ↗
- arxiv
- 2610.01559
Explicit source versions (1)
- v2 · 2026-10-06
- Explicit source version detected · v2
Source date: 2026-10-06 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:id explicit version (official metadata)
- Preprint available
Source date: 2026-10-01 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
World Action Models (WAMs) predict visual futures and robot actions, yet they remain susceptible to task-incomplete imagination, where plausible, action-consistent predictions omit the transition needed for task completion. In this paper, we show that this failure is not inherent to the world model backbone, but emerges when adapted for short-chunk control, which can repeatedly favor plausible local continuations over task-completing transitions. To address this, we introduce Completion Aware Guidance (CAG), a training-free sampling method that guides generation toward task completion. Across representative WAMs, CAG improves success from 64% to 70% on a RoboTwin 2.0 subset and from 69% to 75% in zero-shot simulation, while reducing task-incomplete imagination from 79% to 40%.