Robot Learning from Human Interactions

How robots deployed in the real world can leverage human interactions — through demonstrations, feedback, and natural signals — to learn and improve their abilities over time.

Summary

When robots operate outside the lab, the environment is unconstrained and unpredictable — and people are everywhere. Rather than treating this as a problem, this line of work turns it into an opportunity: humans in the loop as a resource for learning. This includes learning local navigation recovery policies from non-expert demonstrations (Del Duchetto 2018, Faris 2025), adapting social behaviours from implicit interaction signals (Del Duchetto 2022), and developing tools that enable novice users to program and assess robot social abilities. A key challenge is designing principled methods that work with sparse, noisy, and unstructured human input — not carefully curated datasets.

Videos

Projects

References

2025

  1. Modelling robot navigation recovery policies from non-expert users’ demonstrations in-the-wild
    Ahmad Faris, Ayse Kucukyilmaz, Athanasios Polydoros, and 1 more author
    In 2025 International Conference on Robotic Computing and Communication (RoboticCC), 2025

2022

  1. Learning on the Job: Long-Term Behavioural Adaptation in Human-Robot Interactions
    Francesco Del Duchetto, and Marc Hanheide
    IEEE Robotics and Automation Letters, 2022

2020

  1. Automatic Assessment and Learning of Robot Social Abilities
    Francesco Del Duchetto, Paul Baxter, and Marc Hanheide
    In Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 2020
  2. Abstract visual programming of social robots for novice users
    Onis Brown, Laurence Roberts-Elliott, Francesco Del Duchetto, and 2 more authors
    In Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 2020

2018

  1. Do not make the same mistakes again and again: Learning local recovery policies for navigation from human demonstrations
    Francesco Del Duchetto, Ayse Kucukyilmaz, Luca Iocchi, and 1 more author
    IEEE Robotics and Automation Letters, 2018