In Japan, which is an aging society, the number of people with reduced walking ability is expected to increase. Maintaining and improving walking ability is an issue in the rehabilitation field. Therefore, early detection of the risk of falls in the elderly is an important issue. TUG (Timed Up and Go test) is used as one of the evaluation methods for fall risk in many situations. It is thought that TUG can be divided for each movement, measured and analyzed in detail, and indicators for fall risk assessment and walking independence judgment can be defined. However, if it takes time and effort to install a large device or measuring instrument for TUG, the advantage of TUG that it can be easily carried out decreases. By using the IoRT (Internet of Robotic Things) walker developed as gait training device, it is possible to easily and automatically measure the gait of the TUG without any special operation or equipment. In this paper, we propose an automatic classification method for gait sections in TUG test using IoRT walker. The proposed method can detect the walking state using motion information obtained from the walker, and can automatically classify the gait section in the TUG test into ”Sit to stand, Walk 1, Turn Around, Walk 2, Turn and Stand to sit” sections. The effectiveness of the proposed method was confirmed by the TUG Test, which was conducted on various pseudo-impaired gaits.
