English
Gaeilge
Ireland’s over-65 population to double: Why robots could meet the demand for physical therapy
Inside the lab where Ali Al Abbas is teaching rehabilitation robots to care like humans.
Populations are aging, injuries are rising, and the world simply doesn’t have enough clinicians to go around. In Ireland alone, the over-65 population is projected to double by 2051 and 32% of the workforce will be over 50 by 2031.
PhD researcher Ali Al Abbas is engineering a future where a single therapist can teach not just one patient at a time, but dozens of robots at once.
What if a rehabilitation robot could learn an exercise just by watching a video?
Not coded.
Not programmed.
Simply shown.
That’s the idea driving Ali whose current research blends vision science and human-centred design to take on one of healthcare’s most urgent challenges.
An older, more active world is arriving fast and the demand for rehabilitation is rising even faster than the clinicians who can deliver it.
Cobots with a human touch
Ali is clear about the mission:
“We’re bridging the gap between vision-based learning and human-centred care.”
Therapists will be able to teach and monitor multiple patients remotely, safely and efficiently,” says Ali Al Abbas.
Unlike traditional rehabilitation robots, which often require expensive motion-capture systems or ‘kinesthetic teaching’ (where a therapist must physically guide the robot), this framework uses only cameras.
Using vision-based Learning from Demonstration (LfD), the robot watches a therapist perform an exercise and extracts the full 3D motion directly from the video.
That motion is transformed using Dynamic Movement Primitives, giving the robot a smooth, accurate, adaptable memory of the exercise.
The robot doesn’t just copy; it understands the intention behind the movement.
But the real breakthrough is safety.
Safety that feels human
Using Gaussian Mixture Regression (GMR), the system learns the ‘normal’ forces of a healthy limb, building a model of how the exercise should feel.
The robot observes, learns, and then safely guides patients through exercises, adapting to each individual’s comfort level.
During therapy, it continuously monitors interaction forces. If the patient tenses, hesitates, or shows early signs of discomfort, the robot senses it instantly and adjusts, ensuring exercises remain self-paced, personalized, and safe.
Scaling Care Across Clinics
Tested on a UR5e collaborative robot, Ali’s framework has already mastered common upper-limb rehab tasks like shoulder abduction and elbow flexion, performing them with smooth precision and gentle awareness.
Our goal is to make rehabilitation more accessible, scalable, and safe, where one skilled therapist can teach many robots, anywhere.
The implications are significant: One skilled therapist can teach dozens of robots with a single video, scaling expertise across clinics and continents.
This research is funded under RISE@ATU with supervision by Dr Philip Long and Dr Joanne Regan-Moriarty. RISE@ATU is co-funded by the Government of Ireland and the European Union through the ERDF Northern and Western Regional Programme 2021-27.
About Ali Al Abbas
Ali Al Abbas is a robotics researcher at Atlantic Technological University, specialising in rehabilitation robotics and AI. His work focuses on human-robot interaction, language models, and haptic devices for collaborative human-robot systems. He is currently developing an AI-powered robotic system that helps patients do personalised rehabilitation exercises at home. The robot learns how to perform exercises by watching video demonstrations from a therapist and can understand simple language commands. This allows patients to receive guided therapy without needing the therapist to be physically present.
Featured Image: Ali Al Abbas, PhD Researcher at Atlantic Technological University.
For media enquiries, contact:
Jorden McMenamin
Communications Officer
Tel: 074 918 6127
E: jorden.mcmenamin@atu.ie
