World Action Model

ThinkWAMRethinking Latent Futures
for Action Refinement

Propose an action. Rethink its consequences. Refine before execution.

Overview

A future prediction that is not conditioned on the chunk is identical for every proposal and thus carries no proposal-specific signal, limiting its usefulness for refinement before execution.

ThinkWAM conditions a learned residual on the complete action proposal to correct multi-horizon future latents. The same action head uses these corrected futures in a second pass, supported by affordance-guided memory and near-fine, far-coarse prediction. All prediction stays in latent space.

Method

A second action pass, guided by corrected latent futures.

  1. 1Propose

    Use observation–action memory to propose the next action chunk.

  2. 2Rethink

    Correct near, middle, and far latent futures using the proposed action.

  3. 3Refine

    Generate the executable action chunk from the corrected futures.

Architecture details

ThinkWAM uses a proposed action to revise its latent future predictions, then conditions the second action pass on those revised futures. The images used to illustrate latent horizons are conceptual visualizations.

Performance

ThinkWAM against strong baselines across simulation and real-world settings.

Success across six settings

ThinkWAMπ0.5

Loading chart…

ID–OOD generalization

Non-JEPAJEPA-basedThinkWAM

Loading chart…

Inference Efficiency

Simulation Results

Success rate (%) per suite · bold = best, underline = second best.

Simulation latency & action chunks

Simulation Demonstrations

Real-world Results

3 OOD conditions · 30 trials per method and condition

Galaxea A1Z · NVIDIA RTX 4080 SUPER

Real-world Demonstrations

Green-object tidying · Galaxea A1Z

Mechanism Analysis

What the latent correction changes, and where it helps.

Ablation Study

LIBERO and LIBERO-Plus · success rate (%)

† Both action passes are retained; the second pass uses uncorrected futures.

BibTeX

@misc{tang2026thinkwam,
  title  = {{ThinkWAM}: Rethinking Latent Futures for Action Refinement},
  author = {Tang, Haoyun and Deng, Baqiao and Zhou, Enjin and Tian, Songwen
            and Xu, Keyao and Jing, Bowen and Wang, Kun and Ma, Ke
            and Mei, Zhandong and Li, Huibin},
  year   = {2026},
  eprint = {2611.21258},
  archivePrefix = {arXiv},
  primaryClass = {cs.RO},
  url    = {https://arxiv.org/abs/2611.21258}
}