Driving · 2026-10-06
Future Anchored Verification and Online Recovery for World Action Models
Zhibin Qin, Zhenxiong Tan, Xinchao Wang
AI-assisted summary based on the paper. Consult the linked source for methods and limitations. Benchmark numbers appear after editor verification.
TL;DR
FAVOR (Future Anchored Verification and Online Recovery) uses a world action model’s own predicted future frames as anchors to (1) verify execution drift and (2) guide online recovery. An Anchor Verifier compares observations to their anchors (with actions and horizon context) to flag deviations. Anchor-Guided Recovery uses a vision-language model to issue a short corrective instruction to restore the anchored condition, then resumes the task.
Why it matters
For world action models, execution drift can invalidate remaining actions; existing monitors can detect failures but don’t specify what state to restore. FAVOR addresses this by treating the predicted future (that the policy intended to pass through) as both the verification reference and the recovery target, enabling recovery without modifying the base policy.
Method
- FAVOR keeps predicted future frames as anchor states during chunk execution; every Δ steps it verifies the current observation against the corresponding anchor using an Anchor Verifier conditioned on observations, anchors, executed actions, and anchor horizon.
- The Anchor Verifier (using frozen DINOv2 visual tokens fused with action and horizon embeddings, then a GRU and linear head) raises an alarm when its deviation score exceeds a fixed threshold τ.
- When alarmed, Anchor-Guided Recovery uses a VLM to translate the alarmed anchor and current observation into a one-short-instruction corrective command; classifier-free guidance strengthens instruction influence during recovery, and the system resumes once the anchor is restored.
Abstract (from arXiv)
World action models (WAMs) have emerged as a promising paradigm for robotic manipulation. They act by first predicting how a task should be performed and then decoding the actions from that future. However, the remaining actions are invalid once execution drifts from the prediction. Simply replanning from the already out of distribution state rarely restores what the task still requires; existing execution monitors decide when to stop, but not what to restore. We observe that the answer is already in hand: the future the WAM predicted before acting depicts exactly the states it intended to pass through. We introduce FAVOR (Future Anchored Verification and Online Recovery), a lightweight framework that keeps these predicted frames as anchors and uses them for verification and recovery. An Anchor Verifier compares each observation with its anchor, together with the executed actions, to flag deviations that break the task. Anchor-Guided Recovery uses a vision-language model to turn the flagged anchor into a short corrective instruction. Under strengthened instruction guidance, the WAM executes this instruction to return to the intended future. It then resumes the task. FAVOR raises the task success of the base WAM from 97.85% to 98.10% on LIBERO and from 72.60% to 72.98% on LIBERO-Plus without modifying the policy.
Related papers
- Plausible Future, Wrong Choice: Steering World-Action Models toward Intended Future
- What Limits Cross-Task Generalization in World Action Models?
- WAMJET: A Harness for World Action Model Acceleration
- RM-WAM: Reliable Memory-Augmented World Action Models for Long-Horizon Robot Manipulation
- $R^2$-WAM: Repair-and-Reject Post-Training for World Action Models
Publication record
Bibliographic evidence describes where and when a record appeared. It does not validate scientific claims.
First public release: 2026-10-05 · Indexed by Pub-AI: 2026-10-06 UTC
Future Anchored Verification and Online Recovery for World Action Models
Preprint posted: 2026-10-05 · Original record ↗
- arxiv
- 2610.06280
Explicit source versions (1)
- v1 · 2026-10-05
- Preprint available
Source date: 2026-10-05 · First detected: 2026-10-25 UTC · Historical baseline
Source evidence (1)
- Official evidence ↗ · atom:published (official metadata)
Abstract · arxiv
World action models (WAMs) have emerged as a promising paradigm for robotic manipulation. They act by first predicting how a task should be performed and then decoding the actions from that future. However, the remaining actions are invalid once execution drifts from the prediction. Simply replanning from the already out of distribution state rarely restores what the task still requires; existing execution monitors decide when to stop, but not what to restore. We observe that the answer is already in hand: the future the WAM predicted before acting depicts exactly the states it intended to pass through. We introduce FAVOR (Future Anchored Verification and Online Recovery), a lightweight framework that keeps these predicted frames as anchors and uses them for verification and recovery. An Anchor Verifier compares each observation with its anchor, together with the executed actions, to flag deviations that break the task. Anchor-Guided Recovery uses a vision-language model to turn the flagged anchor into a short corrective instruction. Under strengthened instruction guidance, the WAM executes this instruction to return to the intended future. It then resumes the task. FAVOR raises the task success of the base WAM from 97.85% to 98.10% on LIBERO and from 72.60% to 72.98% on LIBERO-Plus without modifying the policy.