Building Trustworthy Autonomy: Human-Centred AI for Robotics

Workshop proposal 1.Workshop Title Building Trustworthy Autonomy: Human-Centred AI for Robotics 2. Organizers • Simona Casini (Corresponding Organizer), University of Pisa, [email protected] • Pietro Ducange, University of Pisa, [email protected] • Fabrizio Ruffini, University of Pisa, [email protected] • Lorenzo Pollini, University of Pisa, [email protected] Note: This Workshop is supported by the project “FAIR - Future Artificial Intelligence Research” - Spoke 1 - funded by the European Union under the NextGenerationEU programme. 3. Workshop Format The workshop will be open for submissions and structured into three main segments: • Invited Talks (1 hour): Three invited speakers will present key research and industry contributions. Each talk will last 15 minutes, followed by 3 minutes of Q&A, for a total of 18 minutes per speaker. • Flash Presentations (30 minutes): A set of 10 selected contributions (from submitted extended abstracts) will be presented through 3-minute talks. • Poster & Demo Session (45 minutes): Following the flash presentations, all contributors will participate in an interactive poster session, where attendees can engage in one-on-one discussions and view demonstrations (if applicable). 4. Workshop Objectives and Main Topics The development of autonomous robotic systems powered by Artificial Intelligence is rapidly transforming the way we interact with machines, environments, and data. As these systems become increasingly embedded in critical domains, from healthcare and agriculture to industry and mobility, trust and human alignment are essential. This workshop aims to create a shared space for researchers, engineers, and practitioners working at the intersection of AI, robotics, and human-centred design. We will explore how to design robotic systems that are not only technically robust and adaptive, but also understandable, transparent, and aligned with human values and goals. The workshop will cover both foundational aspects (such as explainability, uncertainty management, and human-AI interaction) and practical implementations (e.g., robot decision-making in dynamic environments, mission planning, and social interaction). By doing so, we aim to foster discussions on how to move from isolated technical solutions toward integrated architectures that support trustworthy autonomy in real-world scenarios. Topics include (but are not limited to): • Trustworthy and explainable AI for robotic perception, planning, and control • Human-robot interaction and shared autonomy • Learning from human feedback, demonstration, or natural language • Embodied intelligence and awareness in unstructured environments • Task decomposition, mission management, and decision support • Social, ethical, and regulatory dimensions of autonomous robots • Architectures for integrating human-centred AI principles in robotics This workshop also encourages contributions that address the future directions of research and policy around HumanCentred AI and its concrete impact on the design of next-generation robotic systems. 5. Keywords Human-Centred AI, Trustworthy AI, Robotics, Explainability, Human-Robot Interaction, Embodied Intelligence 2025 I-RIM Conference October 17-19, Rome, Italy ISBN: 9788894580570 10.5281/zenodo.17629594 3 6. Preferred Date October 18th (afternoon preferred if possible) 7. List of Speakers • Francesco Marcelloni, University of Pisa - CONFIRMED o Personal page o Foundation FAIR • Giordana Bucchioni, Univeristy of Pisa, AI applications in space robotics – CONFIRMED o Personal page o MUSAPOEM Project • Matteo Razzanelli, Proxima Robotics, Barriers and difficulties of trustworthy AI in industry - CONFIRMED o Personal page o Proxima Robotics 8. Expected Number of Attendees 10 attendees + invited guests 9. Target Audience Researchers, PhD students, engineers, policy makers, and robotics professionals from academia and industry. 10. Workshop Outcomes • I-RIM3D conference proceedings • Video recordings of talks (if permitted) • Networking opportunities for collaborative projects and proposals 11. Plan for Special Issue or Session While we do not currently plan to organize a formal special issue, the workshop is intended as a first step toward building a dedicated community of researchers and practitioners interested in Trustworthy and Human-Centered AI for Robotics. The goal is to establish an open network of collaborators working at the intersection of AI, autonomy, ethics, and humanrobot interaction. To support this objective, we plan to: • Facilitate post-workshop networking, including a shared mailing list and community space for follow-up discussion and idea exchange • Collect interest for future joint activities, such as invited sessions at upcoming conferences (e.g., ICRA, HRI, ...) • Explore the possibility of organizing a focused workshop track or working group • Identify opportunities for collaborative research proposals, particularly in the context of Horizon Europe calls focused on ethical AI, embodied intelligence, or Industry 5.0 This community-driven approach is intended to foster long-term collaboration, support early-career researchers, and align national and European efforts on trustworthy AI in robotics. 12. Call for Extended Abstracts • Regular extended abstracts submitted to I-RIM and associated to the workshop by the program committee based on topic affinity. • A call for extended abstracts (2 pages) via EasyChair will be organised. The CFP will be distributed via mailing lists, social media, and institutional channels. 13. Workshop Webpage A Google site is already designed: https://sites.google.com/unipi.it/hcai-robotics-workshop/home 14. Notes on Registration We will use the two free one-day registrations for the invited speakers or at least for the workshop organizers. 4