Design And Implementation Of Posture Detection Based On Inerial Sensors

Since sitting is the most common posture of modern professional work,maintaining healthy posture in daily work is of great significance to the prevention of disease in the elderly.Appropriate posture,therefore,provides valuable index for healthy.In recent years,the development of micro-mechanical electronic system technology,especially the acceleration and angular velocity sensor,highly promoted the study of posture detection.The attitude detection of single sensor is likely to decrease the accuracy due to the drift of the inertial device and the accumulative error.Therefore,it is the focus of present study to improve the attitude detection accuracy by using multiple sensors for attitude detection and data fusion algorithm.Compared with the human body posture recognition method based on computer vision technology,human gesture recognition and sensor monitoring technology performs better in privacy,accuracy and convenience etc.This paper,based on the status quo of research and development at home and abroad,will deepen the study of sitting posture detection.In this paper,the use of MPU6050 sensors,microcontroller,selector and power supply hardware composed of sitting posture detection system.Velocity data collected through the sensor of three axis acceleration and three axis angular,passed through the I2 C bus protocol to the microcontroller,and then upload to the computer for stored.The original data collected by the MATLAB software are then processed through the method of Kalman filtering.The synthetic processed data then used for extracting feature.Recognition of processed data,by SVM classifier,followed the recognition of different sitting postures.The experiments show that this system is able to discern sitting,forward,hypsokinesis,left or right bank,as well as postures of both legs with different height and weight,and the rate of recognition exceeds 90%.