Correct Motion Advice on Rehabilitation Instruction Robot by Superimposing Instructor CG Model

This paper describes a means for advising a correct motion on a rehabilitation instruction robot that instructs a patient the stretching exercises in rehabilitation. Rehabilitation is one of the ways for people to keep and recover physical functions that are needed in daily life. The rehabilitation is generally performed by a physical therapist or an occupational therapist in a hospital or rehabilitation facility. Recently, the number of therapists has been insufficient because the number of elderly people has increased. We have been studying a robot system to help with the instruction of rehabilitation processes. The instruction robot not only can show a patient how to perform an exercise, but also can evaluate the accuracy of the patient's exercise. In addition, the robot advises the patient how to correct his/her posture if the exercise is incorrect. The rehabilitation instruction robot advises the patient to correct his/her exercises by both the computer-generated voice and the correct exercise shown by the robot. Depending on the contents of rehabilitation exercises, it is difficult to advise a patient in words which body site to correct or how to correct his body site. In this paper, we improve the correct motion advice by superimposing an instructor model created by computer graphics (CG) onto the video image of the patient. Furthermore, the system displays the CG part of the patient body site of the incorrect exercise with a different color. By combining a verbal advice and a visual advice, the system is able to give a patient an easily understandable advice.