Learning a Thousand Tasks in a Day
Pietro Vitiello, Kamil Dreczkowski, Vitalis Vosylius and Edward Johns
Published on the cover of Science Robotics 2025
Mini abstract. We study how different policy components scale with dataset size and data diversity. This led us to the development of a new highly efficient imitation learning paradigm that enables teaching a robot 1000 distinct tasks in just 24 hours.Â
R+X: Retrieval and Execution from Everyday Human Videos
Georgios Papagiannis, Norman Di Palo, Pietro Vitiello and Edward Johns
Published at ICRA 2025
Mini abstract. R+X enables robots to learn skills from long, unlabelled first-person videos of humans performing everyday tasks. Given a language command from a human, R+X first retrieves short video clips containing relevant behaviour, and then conditions an in-context imitation learning technique (KAT) on this behaviour to execute the skill.
One_Shot Imitation Learning: A Pose Estimation Perspective
Pietro Vitiello, Kamil Dreczkowski, and Edward Johns
Published at CoRL 2023
Mini abstract. We study how unseen object pose estimation can be used to transfer a trajectory from a single demonstration to a novel scene. Through a series of investigations, we analyse the key factors influencing this process and showcase the expected performance of state-of-the-art pose estimators on real-world robotics tasks.