Lab-to-Production Workshop

Overview

Recent robotics research increasingly targets physical general intelligence, emphasizing generality, learning, and adaptability across tasks. In contrast, industrial robots deployed in factories and warehouses—numbering in the millions worldwide—are optimized for precision, repeatability, reliability, and safety, but are typically single-purpose and tightly engineered for fixed workflows. This divergence has created a growing gap between what is demonstrated in research labs and what can be deployed in production environments.

This workshop focuses on the challenges of translating recent advances in autonomous manipulation and perception from lab to production. Rather than asking how industry can adopt existing research systems, we emphasize what researchers must change in their methods, assumptions, and evaluation practices to meet industrial constraints. Key research questions include:

• How can general-purpose robot systems achieve industrial levels of accuracy, speed, and robustness?
• What representations, learning paradigms, and system architectures support both adaptability and repeatability?
• How should researchers evaluate and benchmark autonomy under real-world manufacturing constraints?
• What role do explainability, monitoring, and failure handling play in production-ready autonomy?

This workshop aims to unite the industrial robotics community at RSS, bringing together researchers across different domains (e.g., manufacturing, warehousing, logistics) to broaden the discussion on how recent advancements in autonomy and AI have influenced the field. To foster new insights and collaborations, this workshop will feature invited technical talks by renowned experts, spotlight presentations by emerging academic and industrial researchers, and demos by academic and industry research groups to (1) disseminate ground-breaking methods and systems, (2) discuss open challenges facing autonomous industrial perception and manipulation, and (3) surface future research needs in the field.

Discussion Topics

• High-precision, contact-rich assembly and fine motor skills for tight tolerance and dexterous insertion tasks, with perception tightly coupled to control.
• Monitoring, error detection, and failure recovery for robustness during extended assembly sequences.
• Perception and pose estimation under factory realities such as clutter, occlusion, transparent or reflective parts, and high-speed motion.
• Manipulation of diverse objects and with varied dynamics spanning fast pick-and-place of arbitrary objects and manipulating deformables like cables, belts, tape, packaging, and bags.
• Factory world modeling, high fidelity simulation, and benchmarking to support sim2real transfer and evaluate policies for assembly and manipulation.

Invited Speakers

---

Michael Ryoo

Google DeepMind Robotics, Stony Brook University

New York, USA

Talk Title: Robot Learning Research in Academia vs. Frontier AI Labs

Short Bio: Michael Ryoo re-joined Google DeepMind Robotics in March 2026. Prior to that, he was with Salesforce AI Research for two years, and was with the robotics team at Google DeepMind (and formerly Google Brain) for five years. He also holds a tenured position in the Department of Computer Science (CS) at Stony Brook University as an associate professor. Previously, he was an assistant professor at Indiana University Bloomington, and was a staff researcher within the Robotics Section of the NASA's Jet Propulsion Laboratory (JPL). He received his Ph.D. from the University of Texas at Austin in 2008 and B.S. from Korea Advanced Institute of Science and Technology (KAIST) in 2004.

---

Troy Cordie

ARM Hub

Brisbane, Australia

Talk Title: Not Quite Cutting Edge: Robotics Adoption Through a Manufacturer's Lens

Short Bio: Troy Cordie is Director of Industry Research at ARM Hub, where he works with Australian manufacturers to understand and reduce the barriers to robotics adoption. A mechatronics engineer by training, Troy completed a PhD in Modular Reconfigurable Field Robotics in collaboration with the Queensland University of Technology and CSIRO, which led to the development of NeRobot, a modular system designed to simplify bespoke robot deployment in remote and industrial environments. His broader robotics experience spans industrial arms in cold spray 3D printing and computer vision for agricultural robotics. At ARM Hub, he now leads research at the intersection of advanced manufacturing and Industrial AI, working directly with the SMEs who want robotics but have not been able to deploy it.

---

Juxi Leitner

Amazon Robotics

Berlin, Germany

Talk Title: Integrating Robotics, Computer Vision and Artificial Intelligence for Robust Grasping and Manipulation in Real-World Scenarios

Short Bio: Juxi Leitner is an Applied Science Manager at Amazon in Berlin, Germany. He has spent the better part of the last two decades conducting robotics research across academia, government, and industry. Before joining Amazon, Juxi was founder, CEO, and CTO of LYRO Robotics, a startup creating general-purpose pick-and-place robots for the food supply chain based in Melbourne, Australia. In 2017, his team from the Australian Centre for Robotic Vision won the Amazon Robotics Challenge.

---

Nicolas Hudson

Amazon Robotics

Seattle, USA

Talk Title: Robin Pick and Vulcan Stow - Lab to Production

Short Bio: Nicolas Hudson is a Principal Applied Scientist at Amazon Robotics, a position held since 2021. Previously, they served as a Senior Principal Research Scientist at CSIRO's Data61, where they were the Principal Investigator for the DARPA Subterranean Challenge Team and led a team researching robotic capabilities. Nicolas has also worked as a Senior Roboticist at both Google and NASA Jet Propulsion Laboratory, and held engineering roles at X and Boston Dynamics. They earned a Bachelor of Engineering with honors in Mechanical Engineering from the University of Canterbury and a PhD in Mechanical Engineering from Caltech.

Event Schedule

Accepted Abstracts

Call For Papers

We invite extended abstracts of up to 3 pages (excluding references, acknowledgments, limitations, and appendix) formatted in the RSS template and submitted via the L2P Workshop OpenReview console. The extended abstract must be submitted along with references, acknowledgments, limitations, and appendices as one .pdf file. Supplementary videos, websites, or data are welcome in the OpenReview submission but not required, and must maintain author and affiliation anonymity.

Best Extended Abstract Certificate: The workshop will recognize one outstanding contribution with a Best Extended Abstract Certificate.

Submissions will be reviewed in a double-blind process by workshop organizers and attendees. Each submission must nominate one author as a reviewer to evaluate three other contributions, following the RSS main track's reciprocal review model. Authors are strongly encouraged to incorporate the feedback received from the reviewers to be considered for awards. Accepted abstracts will be published on the workshop website, invited for spotlight presentations, and presented as posters.

We encourage forward-thinking submissions of new or in-progress ideas. Workshop paper versions of papers accepted to the RSS 2026 main track or any other prior conferences are not permitted. We strongly encourage in-person participation of at least one author in the workshop. All deadlines are 11:59 PM Anywhere on Earth (AoE).

Rules

• All submissions must adhere to our double-blind review policy.
  • All author names and affiliations must not appear anywhere in the title or text of the submission or supplementary materials.
  • Acknowledgments to people or funding agencies must not appear in the submission.
  • There must be no links to external websites (e.g., YouTube, GitHub, institutional pages) in the submission.
  • All submission metadata must be anonymized.
• All submissions must include the following, which do not count towards the 3-page limit:
• A Limitations section describing assumptions, failure modes, and other limitations with suggestions for future improvements.
• An AI usage statement (even if no AI tools were used).
• All authors must have an OpenReview profile, created at least two weeks before the submission deadline.
• Submissions must address the research questions and topics on this website. Real-robot experiments or convincing evidence of simulation-to-real transfer are encouraged. Non-robotics submissions will be returned without review.
• Papers submitted to the main conference cannot be submitted to this workshop.
• Bot-generated submissions will be desk rejected.
• Accepted submissions must provide poster and video presentations for the website archive by July 10.
• All posters must be size A0 in portrait format.
• Violation of any rules may result in desk rejection.
• In-person presentation is strongly encouraged for at least one author. Remote participation is allowed, but online meeting details will not be provided.

Organizers

Holly Dinkel

Samsung Research America

Bibit Bianchini

University of Pennsylvania

Xiao Liu

Samsung Research America

Kris Hauser

Samsung Research America
University of Illinois Urbana-Champaign

Youngwoon Lee

Yonsei University

Alessio Caporali

University of Bologna

Simon Manschitz

Honda Research Institute

Eric Heiden

NVIDIA

Kelsey Allen

University of British Columbia
Vector Institute for Artificial Intelligence