Short Papers
Up to 4 pages (excluding references, acknowledgements, and appendices), following RSS 2026 format; eligible for poster presentation and potential spotlight talk.
Machine Learning/CORE Unranked
Dex-RSS-26 2026
4th RSS Workshop on Dexterous Manipulation: Scalable Learning for Human-Level Skills
The 4th Workshop on Dexterous Manipulation: Scalable Learning for Human-Level Skills (Dex-RSS-26) at RSS 2026 focuses on advancing dexterous robotic manipulation by leveraging scalable learning from massive video datasets, reliable hardware, and sim-to-real pipelines. It brings together researchers to explore breakthroughs in learning, planning, control, sensing, and hardware design for high-DoF manipulation systems.
Paper fit
Contribution paths
A strong submission should clearly identify its contribution and evaluate it appropriately.
Demos
Video demonstrations of robotic systems submitted via OpenReview; no strict page limit for summary document; encouraged even without formal paper.
Research areas in scope
01
Synergy of Learning, Planning, and Control
What is the right balance between model-free learning and control-theoretic methods?Can we expect a universal foundation policy for daily dexterous manipulation?How can we scale up planning and optimization methods to high-DoF systems?How do we efficiently scale reinforcement learning pipelines across hundreds of tasks without manual intervention?
02
Data, Simulation, and Real-World Transfer
How can we better model complex hand-object contacts and speed up simulation?How do we standardize the collection of human hand data in the wild?How can we develop low-cost, easy-to-use teleoperation systems for large-scale data?How can we make policies robust when training demonstrations are scarce or biased?
03
Perception: Seeing & Feeling the World
How do we ensure policy generalization across varying lighting, clutter, and unseen objects?What is the true role of tactile sensing: complement or equal to vision?What types of tactile sensors are necessary and how can we use them efficiently?
04
Hardware: Redesigning Dexterous Hands
What innovations can make robotic hands more compact, efficient, and low-cost?How can we improve hardware robustness while reducing upkeep costs?What benefits do soft hands or novel compliant mechanisms bring to simplifying manipulation?
05
Dynamic and Forceful Manipulation
How can learning-based policies handle dynamic tasks requiring high-frequency control?How do we build robust policies for forceful, high-speed physical interactions?How can we make high-DoF dexterous hands safely compliant during these interactions?
Policies worth checking twice
- Submission must be double-blind: no author names, affiliations, or identifying acknowledgements in main paper, appendix, or supplementary materials.
- Paper length is limited to 4 pages excluding references, acknowledgements, and appendices.
- Submissions must follow the standard RSS 2026 main conference format.
- Dual submissions are allowed: papers under review at other venues (including RSS 2026 main conference) are welcome.
- Previously published works are permitted if their publication status is explicitly stated at submission.
- Both poster and spotlight presentations must be given in person.
- Demo submissions require a summary PDF and a single zip file containing video demos.
Official sources
Compiled from the official call for papers. The organizers’ pages remain authoritative.
Last verified September 9, 2026