관성센서를 이용한 인체모션 분석에 관한 연구

Abstract

Due to the recent increase of interest in education, welfare, and other recreational activities, various self-coaching systems using motion capture technology are being developed following the demand of the health-care field and development of IT based information technology. To implement self-coaching systems, acquisition of location and direction data on the motions of experts and users necessary for coaching, and analysis of human motion data of motions of experts and users for coaching user motions are necessary. In this thesis, to establish a self-coaching system, a motion capture system using inertial sensors to solve the problems of the previous systems was established, and studies on methods to acquire and analyze data using the established system were done. The inertial sensor based motion capture system is categorized into a 4-joints based motion capture system and a human motion capture system. The 4-joints based motion capture system uses 4 major joints to obtain and express human motion, and are composed of 4 inertial sensors, 3 channel receivers, a PC and a server for establishing a DB. The human motion based motion capture system obtains and expresses human motion, and is composed of 15 inertial sensors, 3 channel receivers, a PC and a server for establishing a DB. Human motion data was acquired from 10 volunteers to obtain and analyze human motion data using each system. The motion data was acquired while the volunteers were listening to a lecture, and to compare this with the normal posture data obtained from experts, this data was categorized as abnormal posture data. The obtained data through the 4-joints based motion capture system is the Euler angle, and this data is analyzed using the right handed coordinate system. According to the analysis results, in the normal posture state, the values of the 12 factor data (Roll, Pitch Yaw of the right shoulder, Roll, Pitch Yaw of the left shoulder, Roll, Pitch Yaw of the right pelvis, Roll, Pitch Yaw of the left pelvis) of the 4 joints average near 0˚, and in the abnormal posture state, this value deviates from 0˚. The data obtained from the human model based motion capture system is in BVH format. This motion data is converted to 3 dimensional location data by the OFFSET stated in the HIERARCHY section. According to the analysis results, the characteristics obtainable by the human model based motion capture system are the Euler angle, 3 dimensional location data, vertical/parallel data of the joints, and the angle inside the display connecting 3 to 4 joint points, and this characteristics data was confirmed as possible to be used to determine normal and abnormal postures.