PC 사용자를 위한 인공지능 기반 거북목 경고 시스템 개발 (Development of a Forward Head Posture Warning System for PC Users by Using Artificial Intelligence Algorithms)

서지정보

· 발행기관 : 제어·로봇·시스템학회
· 수록지 정보 : 제어.로봇.시스템학회 논문지 / 26권 / 5호 / 355 ~ 362페이지
· 저자명 : 정명석, 신동원, 임동원

초록

In recent years, excessive use of laptop computers, tablets, smartphones, and other small portable digital devices has caused many side effects, including the FHP (Forward Head Posture) or the poking chin. It has been reported that FHP results in shoulder pain, neck ache, craniofacial pain, headache, or other adverse conditions. To prevent FHP, it is advised to consciously keep one’s head in an upright posture. This paper proposes a new FHP warning system, which can help a user maintain a good posture by signaling an alarm, when FHP is detected. The system consists of an FHP warning device, a measurement set, and a decision algorithm for FHP. Unlike previous studies and patents, the proposed system delivers the design that is comfortable for daily use (because it is not worn by a user) and highly accurate through the adaptation of an AI (Artificial Intelligence) algorithm. The warning device was created as a turtle doll, which neck and back is elongated and lighted by LEDs, respectively. The measurement set consists of a servo motor and two sensors that read the distances from the computer to the upper face and the chest with the angular displacement of the two parameters. To decide FHP, three popular AI algorithms that are relevant to the classification problem were considered and analyzed, namely the ANN (Artificial Neural Network), SVM (Support Vector Machine), and DT (Decision Tree). Measurement data were collected by 7 different users and 2 chairs with different heights for good posture and FHP scenarios. The test results of the trained algorithms yielded accuracy performances of 73.78%, 83.00%, and 86.26% for ANN, SVM, and DT, respectively. In this study, ANN’s poor performance due to various reasons (e.g., the small training sample size) was discussed. DT, which was suitable for a problem set of 0 or 1 discrete outputs, was selected for the FHP warning system.

영어초록

In recent years, excessive use of laptop computers, tablets, smartphones, and other small portable digital devices has caused many side effects, including the FHP (Forward Head Posture) or the poking chin. It has been reported that FHP results in shoulder pain, neck ache, craniofacial pain, headache, or other adverse conditions. To prevent FHP, it is advised to consciously keep one’s head in an upright posture. This paper proposes a new FHP warning system, which can help a user maintain a good posture by signaling an alarm, when FHP is detected. The system consists of an FHP warning device, a measurement set, and a decision algorithm for FHP. Unlike previous studies and patents, the proposed system delivers the design that is comfortable for daily use (because it is not worn by a user) and highly accurate through the adaptation of an AI (Artificial Intelligence) algorithm. The warning device was created as a turtle doll, which neck and back is elongated and lighted by LEDs, respectively. The measurement set consists of a servo motor and two sensors that read the distances from the computer to the upper face and the chest with the angular displacement of the two parameters. To decide FHP, three popular AI algorithms that are relevant to the classification problem were considered and analyzed, namely the ANN (Artificial Neural Network), SVM (Support Vector Machine), and DT (Decision Tree). Measurement data were collected by 7 different users and 2 chairs with different heights for good posture and FHP scenarios. The test results of the trained algorithms yielded accuracy performances of 73.78%, 83.00%, and 86.26% for ANN, SVM, and DT, respectively. In this study, ANN’s poor performance due to various reasons (e.g., the small training sample size) was discussed. DT, which was suitable for a problem set of 0 or 1 discrete outputs, was selected for the FHP warning system.

참고자료

· 없음