Dr.VCoach: employment of advanced deep learning and human–robot interaction for virtual coaching

Ensuring healthy aging and physical fitness is crucial for leading a happy, independent, and fulfilling life. Daily gentle gymnastic training at home can significantly contribute to maintaining physical health among older adults. Virtual and robotic coaches serve as valuable tools to assist elders during their daily exercise routines. In this paper, we introduce the design of a comprehensive robotic coach capable of guiding patients throughout their training sessions. Additionally, we propose an innovative synthetic dataset generator to train the video action recognition module of the robot effectively. To evaluate the efficacy of our robotic coach, we conducted tests with nine patients aged over 50, who were asked to complete a system usability scale (SUS) survey.