Lifelong Learning and Personalization in Long-Term Human-Robot Interaction (LEAP-HRI)

About
The complex and largely unstructured nature of real-world situations makes it difficult for conventional closed-world robot learning solutions to adapt to such interaction dynamics. These challenges become particularly pronounced in long-term interactions where robots need to go beyond their past learning to continuously evolve with changing environment settings and personalize towards individual user behaviors. In contrast, open-world learning embraces the complexity and unpredictability of the real world, enabling robots to be “lifelong learners” that continuously acquire new knowledge and navigate novel challenges, making them more context-aware while intuitively engaging the users.

The fourth edition of the “Lifelong Learning and Personalization in Long-Term Human-Robot Interaction (LEAP-HRI)” workshop seeks to bring together interdisciplinary perspectives on real-world applications in human-robot interaction (HRI), including education, rehabilitation, elderly care, service, and companion robotics. The primary objective is to explore the concept of lifelong robot learning and the ability to continually adapt to users, contexts and environments in long-term HRI. The workshop’s goal is to foster collaboration and understanding across diverse scientific communities. This will be achieved through a combination of invited keynote presentations and in-depth discussions facilitated by contributed talks, break-out session, and a debate centered around the workshop’s theme. Aligned with the theme of the HRI 2024 conference, “HRI in the real world”, our workshop adopts the theme of “open-world learning”. This encourages the exploration of HRI theories, methodologies, designs, and studies focused on lifelong learning and personalization to learn and adapt to new concepts, users, and tasks after deployment in real-world applications in HRI.

Schedule

Time (GMT-6)
09:00 am - 09:10 am		Organizers Introductory Remarks
09:10 am - 09:40 am		Sonia Chernova Personalized Robot Assistance in the Home: Unobtrusive Adaptation to User Habits Abstract Human homes present one of the greatest challenges for autonomous robotic systems as no two homes, or two users, are exactly alike. This talk will explore two ways in which robots can provide assistance to users in the home in an unobtrusive way, without explicit instructions from the user. We will discuss techniques by which context clues from the user’s environment can be used to personalize robot actions to match diverse user habits.
09:40 am - 10:00 am		Contributed Talks 1 1. Designing Long-Term Interaction for Robot-Assisted Recovery after Critical Injury 2. Prosody for Intuitive Robotic Interface Design: It's Not What You Said, It's How You Said It
10:00 am - 10:30 am		Coffee break
10:30 am - 11:00 am		Silvia Rossi Proactive Behavior and Theory of Mind for Behavioral Change Abstract TBD
11:00 am - 11:30 am		Contributed Talks 2 1. Federated Learning of Socially Appropriate Agent Behaviours in Simulated Home Environments 2. Train your robot in AR: investigating user experience with a continuously learning robot 3. Toward Measuring the Effect of Robot Competency on Human Kinesthetic Feedback in Long-Term Task Learning 4. Towards Learning Interpretable Features from Interventions
11:30 am - 12:30 pm		Debate: Tony Belpaeme (moderator), Georgia Chalvatzaki (debater), Hae Won Park (debater), Siddhartha Srinivasa (debater)
12:30 pm - 12:50 pm		Breakout Session
12:50 pm - 01:00 pm		Organizers Concluding Remarks

Papers

Designing Long-Term Interaction for Robot-Assisted Recovery after Critical Injury [link]
Carl Bettosi, Lynne Baillie and Susan Shenkin

Prosody for Intuitive Robotic Interface Design: It's Not What You Said, It's How You Said It [link]
Elaheh Sanoubari, Atil Iscen, Leila Takayama, Stefano Saliceti, Corbin Cunningham and Ken Caluwaerts

Federated Learning of Socially Appropriate Agent Behaviours in Simulated Home Environments [link]
Saksham Checker, Nikhil Churamani and Hatice Gunes

Train your robot in AR: investigating user experience with a continuously learning robot [link]
Anna Belardinelli, Chao Wang, Daniel Tanneberg, Stephan Hasler and Michael Gienger

Toward Measuring the Effect of Robot Competency on Human Kinesthetic Feedback in Long-Term Task Learning [link]
Shuangge Wang, Brian Scassellati and Tesca Fitzgerald

Towards Learning Interpretable Features from Interventions [link]
Erin Hedlund-Botti, Nina Moorman, Julianna Schalkwyk, Chuxuan Yang, Lakshmi Seelam, Sanne van Waveren, Russell Perkins, Paul Robinette and Matthew Gombolay