Body Attitude Monitoring System Design For The Elderly Based On Inertial Sensors

Today's society is suffering from the problem of aging population.It is not rarely seen in reports nowadays that the elderly are more likely to become victims of abnormal situations such as tripping down or prolonged sitting due to the deterioration of their physical conditions.Along with the development of computer technology and internet communication,personalized medical services have made considerable progress,enabling more people to consider home-based and community-based nursing as an option.Therefore,it is of great social significance and practical value to design and develop a body posture monitoring system with functions of remote monitoring,abnormal state alarming and locating.Body posture monitoring system can be divided into video-based monitoring system and wearable sensor-based monitoring system according to their characteristic parameters.Video-based monitoring system captures original image through camera and locates human joint points through methods such as machine learning and deep learning in order to recognize body posture.Such methods might lead to problems like personal information leakage,huge amount of calculation or high cost.And their application scenarios are mostly confined indoors.Wearable sensor-based monitoring system embeds micro sensors in wearable equipment.By collecting the information like acceleration and angular velocity generated during body motion,it could compare the data with a pre-set threshold value and recognize posture.Such methods do not expose personal information,have low cost and low computational complexity,and have no restrictions on application scenarios.Inertial sensors are sensors for detecting and measuring acceleration,tilt,impact,vibration,rotation and multiple degrees of freedom motions.They are widely used in navigation and positioning,human-computer interaction,posture measurement and other fields.Among them,the automatic and heading reference system based on accelerometer,gyroscope and magnetometer can not only provide accurate posture and navigation information for aircraft,but also be widely used in wearable intelligent devices to assist the progress of posture monitoring and so on.Based on the background mention above,in this paper we designed and implemented a body posture monitoring system based on inertial sensors.The system consists of three parts:posture monitoring terminal,server and supervising terminal.Not only can it detect the fallen state of human body and alarm in time,but also monitor the posture and position information of human body remotely and count the number of steps.In addition,this system implements two-way voice interaction function.The posture monitoring terminal can download voice messages from the server side and play,record vocal information and upload them to the server,which provides a convenient way for the elderly to communicate with their caregivers.The main work of this paper is as follows:(1)Based on the traditional C/S structure model,this paper adopts the three layers structure model of posture monitoring terminal,server and supervising terminal,which could reduce the communication cost and improve the stability and scalability of the system through proper assignment of system tasks.(2)Designed and implemented the posture monitoring terminal.The terminal will be worn on the waist of the user and be powered by lithium battery,which is portable and has low power consumption.The terminal is embedded with inertial measurement unit composed of accelerometer,gyroscope and magnetometer,which can not only accomplish human posture recognition,but also have the function of location and voice interaction.(3)Based on the motion characteristics of the elderly,a simple and flexible human posture recognition algorithm is developed,which could acquire posture angle based on quaternion method and recognize human posture by combining various kinematics features.Tests show that the algorithm can accurately and quickly recognize human posture,and the recognition rate reaches 90%.(4)Designed and implemented a step-counting algorithm based on dynamic threshold.The algorithm is based on the acceleration information generated during human motion.It extracts features and calculates threshold with pre-set time window,which has high accuracy and good robustness.(5)Built an HTTP server based on QT,which can display the user's posture information,location information and steps under the GUI interface,and record the data into the database for query from supervising terminal.