IROS 2025 Workshop - Learning From Teleoperation

Robot Learning From Human Teleoperation:

Prospects and Challenges

IROS 2025 - Workshop

October 20th - Afternoon Session - Pressroom D

The objective of this workshop is to connect an audience from diverse organizations actively working on Learning from Demonstrations (LfD) for robotic manipulation, specifically leveraging robotic teleoperation technologies.

Overview

Imitation Learning (IL) from human teleoperation has emerged as a powerful approach for training robotic systems in complex manipulation tasks. However, challenges remain in developing effective strategies for data collection, ensuring standardization across diverse teleoperation setups, and integrating tactile sensing for improved learning from demonstration (LfD). This workshop will bring together leading researchers and practitioners to explore these key challenges, addressing the question: What are the best strategies to collect and standardize teleoperation data, and what is the role of tactile sensing and haptic feedback in LfD?

Featuring expert speakers, the workshop will present insights on best practices for designing teleoperation interfaces, curating high-quality datasets, and leveraging tactile feedback for enhanced imitation learning. A poster session, complemented by selected lightning talks, will provide an interactive platform for participants to share innovative research, fostering collaboration and discussion on the future of LfD in robotics.

Poster Submission

Invited Speakers

Keynote 1: "Learning Keypoint-based Task Representation from Human Demonstrations for Manipulation Tasks"

Keynote 2: "Dexterous Tactile and non-tactile telerobots for human demonstrations and data capture - some experiences"

Asfour, Tamim

Professor at the Institute for Anthropomatics and Robotics (IAR) at the Faculty of Computer Science at the Karlsruhe Institute of Technology (KIT), Germany. Since 2012, he has headed the Chair of High-Performance Humanoid Technologies.

Walker, Rich

Director of Shadow Robot Company, Director of ADRA Association, Director of euRobotics aisbl

Keynote 3: "ergoCub: Embodied AI from Computational co-design and Imitation Learning"

Keynote 4: "Leveraging (imperfect) teleoperation for dexterous robotic manipulation"

Pucci, Daniele

Head of Artificial and Mechanical Intelligence Researcher at IIT. Since September 2021, Daniele is the PI leading the Artificial and Mechanical Intelligence research line at IIT.