Full Papers
Up to 8 pages in NeurIPS or ICLR format, with potentially large-scale experiments; submitted through OpenReview; accepted papers are non-archival and may be submitted elsewhere.
Machine Learning/CORE Unranked
NeurIPS 2026
Robot Learning with World Models: Capabilities, Frontiers, and Challenges
The Robot Learning with World Models Workshop at NeurIPS 2026 brings together researchers to advance Physical AI by exploring how world models can improve robot learning, reasoning, and control. The workshop focuses on bridging the gap between simulated dynamics and real-world physical interactions, with emphasis on vision-based world models, multi-modal sensing, physical accuracy, and evaluation benchmarks.
Key deadlines
Verified September 9, 2026Full paper
September 3, 2026
AoE
Conference timeline
Submission and decisions
Full paperKey deadline
September 3, 2026 · AoE
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Short Papers
2-4 pages in NeurIPS or ICLR format, with proof-of-concept demonstrations; submitted through OpenReview; accepted papers are non-archival and may be submitted elsewhere.
Demo and Networking Proposals
1-page proposal via Google Form; includes description of what to show or discuss and optionally a demo; accepted proposals are presented in the Demo and Networking session.
Research areas in scope
01
Topics of Interest
World Action Models (WAMs): Unifying world dynamics, spatiotemporal understanding, and robot action generation.Learning, reasoning, and evaluation with imagined rollouts: Leveraging latent/observation space rollouts for planning, causal reasoning, and safety.Multi-modality beyond vision: Integrating tactile sensing, proprioception, force feedback, audio, and other modalities.Physical accuracy and spatiotemporal consistency: Simulating stable reality, realistic dynamics, and reliable control.Evaluation metrics and benchmarks: Action-conditioned metrics and standardized benchmarks focusing on physical plausibility.
Policies worth checking twice
- All accepted papers are non-archival and may be submitted to archival conferences or journals.
- Full papers must be up to 8 pages, short papers 2-4 pages, excluding references and appendices.
- Camera-ready submissions must be de-anonymized and follow the NeurIPS style template.
- Demo and networking proposals are submitted via a Google Form, not OpenReview.
- Submissions must be made through OpenReview for papers and via Google Form for demos.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026