Robots for Logistics in Social Environments


There is a strong need for robots capable of efficiently handling tasks in workspaces shared with humans. As a matter of fact, recent progress in robot safety has enabled the development of robots capable of sharing their workspace with humans without endangering them. But safety has come at the cost of efficiency, and such robots have seen limited success as a result: beyond safety, human trust and joint productivity are still open challenges.
The objectives of this chair are to provide robots with the decisional capacity to navigate in shared environments not only safely, but also efficiently through nonverbal cooperation and, ideally, without disturbing the social environment. We aim more specifically at logistics scenarios, where robots need to have significant speed and inertia to be useful, making these issues crucial and significantly more demanding than in the existing literature where collaborative robots usually have reduced speed and inertia.
Our strategy is to develop a layered approach, addressing motion safety with traditional symbolic AI (automated planning and constraint-based reasoning), nonverbal cooperation with Reinforcement Learning (RL) and social integration with Imitation Learning (IL).



Published on  January 11, 2024
Updated on January 11, 2024