>> Title of the talk: User-Centred Machine Learning for Assistive Robotics
Abstract:
Assistive robots can greatly enhance the independence and quality of life of those most in need. However, developing intelligent robotic systems that learn to help and adapt to users remains challenging. Unlike other AI applications, assistive robotics faces data scarcity due to the high cost of real-world data collection and the need for human-in-the-loop learning and personalisation. As personal robots are deployed in homes, vast amounts of sensitive interaction data (such as gaze, facial expressions, and environmental context) will become available. Effectively leveraging this data requires machine learning strategies that integrate the user while protecting their privacy. In this seminar, we will explore techniques such as continual and federated learning, along with learning from demonstration, to help robots acquire complex skills and adapt over time. We will also discuss how sensor data can improve human-robot understanding, particularly through multimodal fusion for intention prediction, shared control, and assessing trust in the robot. Real-world applications, including robotic wheelchairs, will be used to illustrate the challenges and benefits of user-centred learning for assistive robotics.
>> If you have any questions, please don't hesitate to contact me,
Regards
-------------------------------------------------------
Dr. Amir Aly, PhD, FHEA, MBCS, SIEEE
Lecturer in Artificial Intelligence and Robotics
Programme Manager of Artificial Intelligence
UK and Ireland IEEE-RAS Chapter Vice Chair
School of Engineering, Computing, and Mathematics
Room A307 Portland Square, Drake Circus, PL4 8AA
University of Plymouth, UK