Designing and Evaluating Posture Recognition Algorithm on Temi Robot in Reaktor's Office: A Pilot Study with AI Hub Tampere

If office employees sit for long hours at a desk or in front of their computer, they are likely to strain their muscles and cause tension in their neck, shoulders, and back. If this persists for a long period, health issues, including neck and back pain, headaches, muscle pain, etc., can develop. By fixing their posture and adopting ergonomic practices, as well as having active breaks, office employees can reduce the risk of injury and discomfort and improve their overall health and well-being.

The CloudSEA.AI research group at Tampere University has developed a posture recognition algorithm that was integrated with an autonomous robot, Temi. AI Hub Tampere collaborated with CloudSEA.AI and Reaktor to design and evaluate the “wellness robot” Temi in Reaktor’s Tampere office.

The purpose of the office wellness robot is to move around the office premises, take pictures of the employees’ sitting posture at certain intervals, and send them to a stateless server to run the posture recognition algorithm and analyze the posture of the office employees. The goal of our study was to understand how to provide acceptable feedback to motivate employees to undertake wellness interventions and correct their posture for better ergonomics in the office premises. With this technology in place, we aimed to provide feedback to the employees in a manner that would encourage them to improve their well-being. We conducted two contextual inquiries and one co-design workshop to understand the users’ needs and expectations from the robot. We then designed and implemented our concept on Temi robot to conduct an evaluation study in an IT office environment to understand the acceptance of the robot.

Based on the data gathered from contextual inquiries and co-design workshop, we designed two roles for the wellness robot. The first role was playful and friendly (Wellness Paparazzi), while the second personality was calm and authoritative (Wellness Police). After the evaluation, the employees preferred the playful but calm personality for the robot in the office environment. We found that this personality helped to create a comfortable and engaging environment for the employees. We presented our results to Reaktor and CloudSEA.AI, and according to them, it was interesting to learn how the employees were reacting to the robot and the feedback we received regarding the concept. Reaktor gave us a rating of 4 out of 5 for the pilot study, and we hope to collaborate again soon for a second pilot.

Written by Aparajita Chowdhury

Keep reading