Learning from human feedback
This line of work explores ways robots deployed "in the wild" can learn leverage the human presence for learning to get better over time, improving quality of human-robot interaction and the robot's autonomy.
Summary
When we take our robots outside the labs, due to the unconstrained nature of real-world deployments where the environment is unknown, always changing and contains people (whose behaviour is typically unpredictable), designing robots that can function autonomously becomes a very challenging task. However, the presence of humans in these enviornments provide a great opportunity for robots to learn how to improve their abilities —for example, on how to become more engaging to users during guided tours in a museum (Del Duchetto 2022)— and how to correct errors —such as, improving navigation from human demonstrated recovery trajectories (Del Duchetto 2018). I am interested in creating interfaces of communication between robots and their users to enable natural interactions and robot learning in situated interactions.
Videos
Projects
- Lindsey: the Tour Guide Robot, Funded by Lincolnshire County Council, Partners: University of Lincoln (PI: Prof Marc Hanheide). Link: https://lcas.lincoln.ac.uk/wp/projects/lindsey-a-robot-tour-guide/
- Trustworthy Accessible Robots for Inclusive Cultural experienceS (TARICS), Funded by UKRI TAS Hub, Partners: University of Lincoln (PI: Maria Galvez Trigo), University of Nottingham, Lincolnshire County Council, NICER Group - Nottingham International Consortium for Educational Research. Link: https://tas.ac.uk/research-projects-2022-23/tarics/
References
2022
2020
- Automatic Assessment and Learning of Robot Social AbilitiesIn Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 2020
- Abstract visual programming of social robots for novice usersIn Companion of the 2020 ACM/IEEE International Conference on Human-Robot Interaction, 2020
2019
- Lindsey the Tour Guide Robot-Usage Patterns in a Museum Long-Term DeploymentIn IEEE International Conference on Robot & Human Interactive Communication (RO-MAN), 2019