Content

Topics

Over the past five years, the organizers have established a series of workshops on deformable object manipulation at IROS and ICRA that have been a central venue for discussing progress, challenges, and opportunities in the rapidly growing field of deformable object manipulation. Building on this foundation, the DOM-R3 workshop introduces a renewed perspective that explicitly emphasizes system-level challenges, including evaluation protocols, benchmarking practices, reproducibility, and the transition from laboratory demonstrations to real-world robotic applications. The workshop will solicit contributions spanning both fundamental research and practical system-level challenges. In addition to technical presentations, the workshop will promote discussion on reproducibility and experimental methodology, encouraging submissions that analyze benchmarking practices, evaluation protocols, and lessons learned when reproducing or deploying deformable manipulation systems across different platforms. To further stimulate community discussion, the workshop will feature a panel discussion on reproducibility and benchmarking in deformable manipulation, bringing together researchers to discuss experimental standards, dataset sharing, and evaluation methodologies. As an additional perspective on real-world system performance, the workshop will include a presentation from the winning team of the recent Garment Manipulation Challenge (https://lehome-challenge.github.io/), highlighting insights from both simulation and real-world tasks. The workshop will reflect the state of the art by engaging with emerging research directions, including vision–language–action models, foundation models for robotics, reinforcement and imitation learning, dexterous and bimanual manipulation, and scalable simulation tools. The invited speakers and selected contributions will represent current advances across these areas, providing a comprehensive view of modern approaches to deformable object manipulation. The scope of the contributed papers includes:

• Representation and state estimation
• Simulation and modeling
• Transfer from simulation to reality
• Learning to manipulate using data-driven methods such as reinforcement learning and learning from demonstrations
• Perception: state tracking, parameter identification, property detection (e.g. landmarks for garments) and classification, etc.
• Control, visual servoing and planning
• Use of foundation models, such as large vision and language models, and associated large datasets
• Specialized tools, e.g. grippers, and sensors

Workshop format

The workshop will involve the following sessions to maximize engagement and foster exchange of ideas:

• Invited Talks: Leading academic and industrial speakers will give 20-minute presentations with 5 minutes of Q&A.;<\li>
• Poster Session for Contributed extended abstracts: Authors may submit 3-page extended abstracts (unlimited references and appendix).
• Research Spotlight Talks: A subset of contributed papers will be selected for short spotlight presentations. These talks will emphasize emerging directions, early results, and ongoing work.
• Competition Spotlights: Winners of the LeRobot Garment Manipulation Challenge will present “lessons learned,” focusing on design choices, challenges, and successes in perception, modeling, and control.
• Panel Discussion on Reproducibility: The workshop will conclude with an interactive panel discussion involving invited speakers, organizers, and community members. The panel will focus on reproducibility and benchmarking in deformable manipulation, addressing topics such as evaluation protocols, dataset sharing, experimental standards, and the challenges of comparing results across different robotic platforms. Audience participation will be actively encouraged through open questions and moderated discussion.