Name Truong Cong Thang Class 19ECE Wearable device to detect and warn of falls for the elderly Truong Cong Thang Da Nang University of Science and Technology Department of Electronic and Communicaion[.]
Name: Truong Cong Thang Class: 19ECE Wearable device to detect and warn of falls for the elderly Truong Cong Thang Da Nang University of Science and Technology Department of Electronic and Communicaion Engineering Hoa Khanh Bac, Lien Chieu, Da Nang, Viet Nam Email:truongcongthangran@gmail.com Abstract— Most elderly people will be alone at home If they fall, it is very difficult to call for help and this will be very dangerous to their health or even their lives The main purpose of this report is to design a device capable of detecting falls and alerting family members In this study, we use Sensor module ADXL and microcontroller F303RE, this system uses 3dimensional accelerometer xyz and is designed to be wearable When detecting an elderly person's fall, the system will sound an alarm, and at the same time will send text messages, GPS positioning, and call family members The study was based on four types of activity: walking, standing, exercise, and falling Keywords— Sensor module ADXL, F303RE, text messages, GPS positioning 1,STM32 microcontroller STM32 is one of ST's popular chip lines with many common families such as F0, F1, F2, F3, F4… Stm32f303 belongs to the F3 family with ARM COTEX M4 core STM32F303 is a 32 bit microcontroller, the maximum speed is 72Mhz The charging circuit as well as the programming tools are quite diverse and easy to use microcontroller I INTRODUCTION In Vietnam, there are many elderly people who stay at home alone while their children go to work This leads to the fact that if the elderly fall, it will be difficult to call for help and seek help from others Falling is a common problem among the elderly because their physical strength gradually weakens 33% of elderly people fall one or more times per year [3] This will lead to the risk of physical and mental injury, especially can lead to death[4][5] Fig 1: STM32F303RE nucleon microcontroller Some of the main applications: used for drivers to control applications, control common applications, handheld devices, computers and peripherals, basic GPS, industrial applications, programming equipment PLC, inverter, printer, scanner, alarm system, intercom device It can be replaced by Arduino Nano Microcontroller, this board is suitable for embedded devices [6] 2, Sensor module ADXL Currently, in Vietnam, fall warning and detection systems and devices are not yet widely used This is because very few people think about elderly people falling until it actually happens and cause serious consequences if not detected in time The fact that we fall, especially the elderly, will cause injuries and will increase the cost of long-term treatment[3], In this study, we use accelerometer, to determine the acceleration of axes xyz and microcontroller In case the elderly fall, the system will issue a warning through the horn module, play a sound and at the same time send a message, GPS positioning and call to relatives in the house That way, we will limit the danger of falling when detected and handled in time Fig 2: Sensor ADXL 345 ADXL345 is a small size 3-axis accelerometer module , low power consumption, high resolution 3, II MATERIALS AND METHOD A Systems architecture In general, the fall detection system consists of STM32F303RE Nucleon , ADXL Accelerometer Sensor, SIM800L Module, 2A 5V Charge and Discharge Integrated Circuit, 3.7V 4.2V 18650 Lithium Battery Charge Booster Board, Whistle module The block diagram and the schematic diagram of the peripheral circuit and the source circuit are shown in Fig and Fig respectively Module SIM800L Fig 3: Module SIM800L Inheriting the functions from previous generations of sim modules, the GSM Sim 800L Module has the ability to send SMS messages, listen, call, like a phone, but has the smallest size in all types of SIM modules (25 mm x 22 mm) 4, 2A 5V Charge and Discharge Integrated Circuit, 3.7V 4.2V 18650 Lithium Battery Charge Booster Board C Fall Detection Algorithm and Flow Diagram Similar to a prior work [7], the parameters utilized to calculate the total vector acceleration and angular velocity The equation can be used to compute the signal vector magnitude (SVM), also known as the total acceleration vector: K2 = gx2 + gy2 + gz2 Where gx, gy, and gz are the gravitational acceleration of the x-axis, y-axis and z-axis 1, Flowchart of the system Start Fig 4: Charging circuit Device Configuration Used to charge the entire device, is an indispensable part as well as contributes to making the device work 5, Whistle module Get value from ADXL345 sensor Calculation of axes acceleration Module Sim receives data, sends SMS and makes calls Fig 5: Whistle module and button During the siren sound, there is an additional button used, if someone hears and detects it, press the button to turn off the siren Finish Fig 8: Fall detection Flowchart First of all, when starting, the device will be started and accordingly when receiving a signal from the sensor, the system will calculate according to the threshold we have measured, if it is exceeded, it will receive data to send SMS and finish 2, Flowchart of fall detection algorithm B Block diagram and schematic diagram of Fall Detection Sensor block Control block Communication block Warning block Power block Start Preset threshold value “a” according to survey range Fig 6: Block diagram Main function of blocks Control block: controls the operation of the system Power block: supply power to the operating system Sensor block: provides angular acceleration value Communication block: transmit and receive data with Microcontroller to send SMS Alarm block: receive Microcontroller data to warn via buzzer Read gx,gy,gz values from ADXL345 K2 = gx2 + gy2 + gz2 K2 > a2 Notifications via call, SMS Fig : Flowchart of fall detection algorithm When starting the device, we will preset a specified acceleration threshold for falls, which will be explained in the next section After having read 3D data from ADXL345 sensor, based on the given formula there are cases happening: In the first case, if k exceeds the specified threshold a, this will result in a signal being sent to the communication block, sending a call signal and sending an SMS After the Fig 7: Schematic diagram By looking at the schematic we can draw the circuit, know how to wire and build the circuit It represents the connection of the component's pins elderly person is rescued, placed on a shelf or upright, this will result in the accelerometer working again and the same procedure In the second case, if k is less than the specified threshold a, this will result in a return to the sensor accelerometer, and the procedure is the same as for the one case III EXPERIMENTAL RESULTS Simple motion detection method: Just setting thresholds for each axis of acceleration will not be reliable, because with different fall directions each axis has different acceleration Therefore, to determine the large acceleration of the fall phenomenon, we consider the absolute acceleration of the object, that is, simultaneously investigate the accelerations of the axes and calculate the common acceleration by the formula: K2 = gx2 + gy2 + gz2 Calculate the acceleration of walking, standing up and sitting down, exercise and falls by using STMstudio software: Fig 12: Acceleration of movement during exercise When exercising, it is not more than 3g and only fluctuates at 1-1.5g Fig 13: Acceleration when falling On the graph, we see that the acceleration when falling exceeds 3g Therefore, the threshold value when falling will be near or above the value of 3g, depending on whether the fall is light or strong, the value will change Fig 10: Acceleration of commuting motion Experimental results: Numbe Sample r We see that normal walking motion acceleration always fluctuates at a level not exceeding 1.5g And usually varies in the range of 0g -0.7g Fig 11: Acceleration of standing up and sitting down motion Observing the graph in the case of standing up and sitting down, we see that there will be a higher peak than the rest, but with a normal sitting posture, the acceleration does not exceed 3g A person walking normally and sitting down will have an increase in acceleration, then a steady acceleration of around 1g while sitting, followed by an increase when standing up Commutin g motion Standing up and sitting down motion Movement during exercise Falling Time s Accuracy( %) 50 Identificatio n Fall Do not fall 44 50 47 94% 50 10 40 80% 50 40 10 80% 88% TABLE 1: Identification test results IV.CONCLUSION AND FUTURE WORK In this paper, We design a wearable device to detect and warn the elderly to fall by motions The results show that the probability and sensitivity of the system are quite stable This system was developed for the elderly who stay at home alone and ensure their safety, limiting the occurrence of unwanted incidents As for the advantages, it is easy to wear back and the cost of original components used for research is cheap and widely used Additional improvements will include a more precise fall detection algorithm, a better attachment mechanism, and a lighter and smaller fall sensor Yang, Zhaohui Shen, and Dong Xuan, “PerFallD: A Pervasive Fall Detection System Using Mobile Phones,” Journal of computers, 2010 [1] World Population Ageing Report 2015 Available: http://www.un.org [2] Statistik Penduduk Lanjut Usia 2014 Available: http:// www.bappenas.go.i d [3] Yujia Ge and Bin Xu, “Detecting Falls Using We will continue to work on perfecting a more stable, smoother system with a smaller size It is possible to combine more AI to be more accurate in fall data, combine more Web sites so that loved ones can monitor their health status REFERENCES [1] World Population Ageing Report 2015 Available: http://www.un.org [2] Statistik Penduduk Lanjut Usia 2014 Available: http:// www.bappenas.go.i d [3] Yujia Ge and Bin Xu, “Detecting Falls Using Accelerometer by Adaptive Thresholds in Mobile Devices,” Journal of computers, 2014 [4] Jiangpeng Dai, Xiaole Bai, Zhimin Accelerometer by Adaptive Thresholds in Mobile Devices,” Journal of computers, 2014 [4] Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, and Dong Xuan, “PerFallD: A Pervasive Fall Detection System Using Mobile Phones,” Journal of computers, 2010 [1] World Population Ageing Report 2015 Available: http://www.un.org [2] Statistik Penduduk Lanjut Usia 2014 Available: http:// www.bappenas.go.i d [3] Yujia Ge and Bin Xu, “Detecting Falls Using Accelerometer by Adaptive Thresholds in Mobile Devices,” Journal of computers, 2014 [4] Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, and Dong Xuan, “PerFallD: A Pervasive Fall Detection System Using Mobile Phones,” Journal of computers, 2010 [ ] M K K a r l s s o n , H M a g n u s s o n , T v o n S c h e w e l o v , a n d B E Rosengren, “Prevention of falls in the elderly —a review,” Osteoporosis Internationa l, v o l , n o ,pp.– , [] T.Shany,S.J Redmond, M.R.Naraya nan,andN.H Lovell, “Sensors-based wearable systems for monitoring of human movement and falls,” IEEE Sensors Journa l, v o l , no.,pp – , [ ] M K K a r l s s o n , H M a g n u s s o n , T v o n S c h e w e l o v , a n d B E Rosengren, “Prevention of falls in the elderly —a review,” Osteoporosis Internationa l, v o l , n o ,pp.– , [] T.Shany,S.J Redmond, M.R.Naraya nan,andN.H Lovell, “Sensors-based wearable systems for monitoring of human movement and falls,” IEEE Sensors Journa l, v o l , no.,pp – , [ ] M K K a r l s s o n , H M a g n u s s o n , T v o n S c h e w e l o v , a n d B E Rosengren, “Prevention of falls in the elderly —a review,” Osteoporosis Internationa l, v o l , n o ,pp.– , [] T.Shany,S.J Redmond, M.R.Naraya nan,andN.H Lovell, “Sensors-based wearable systems for monitoring of human movement and falls,” IEEE Sensors Journa l, v o l , no.,pp – , [ ] M K K a r l s s o n , H M a g n u s s o n , T v o n S c h e w e l o v , a n d B E Rosengren, “Prevention of falls in the elderly —a review,” Osteoporosis Internationa l, v o l , n o ,pp.– , [] T.Shany,S.J Redmond, M.R.Naraya nan,andN.H Lovell, “Sensors-based wearable systems for monitoring of human movement and falls,” IEEE Sensors Journa l, v o l , no.,pp – , Ghassemian, and Binh Q Tran, “Optimization of an Accelerometer and Gyroscope-Based Fall Detection Algorithm,” Journal of Sensors, Vol 2015, 2015 [1] Zhenyu Chen, Yiqiang Chen, Lisha Hu, Shuangquan Wang, Xinlong Jiang, “Leveraging Two-Stage Weighted ELM for Multimodal Wearables Based Fall Detection” Springer Proceedings of ELM-2014 Volume 2, Proceedings in Adaptation, Learning and Optimization Volume 4, Pages 161-168, 2015 [2] J Surana, C.S Hemalatha, V Vaidehi, S.A Palavesam, M.J.A Khan, “Adaptive learning based human activity and fall detection using fuzzy frequent pattern mining”, Recent Trends in Information Technology (ICRTIT), 2013 International Conference, Pages 744-749, July 2013 [3] Yujia Ge and Bin Xu, “Detecting Falls Using Accelerometer by Adaptive Thresholds in Mobile Devices,” Journal of computers, 2014 [4] Jiangpeng Dai, Xiaole Bai, Zhimin Yang, Zhaohui Shen, and Dong Xuan, “PerFallD: A Pervasive Fall Detection System Using Mobile Phones,” Journal of computers, 2010 [5] Shumway-Cook, A Ciol, MA Hoffman, J Dudgeon, BJ Yorkston and K Chan,“Falls in the Medicare population: incidence, associated factors, and impacton health care,” Phys, 2009 [6] Arduino Nano Module Available: https://www.arduino.cc/en/uploads/Main/ ArduinoNanoManual23.pdf [7] Huynh, Quoc T., Uyen D Nguyen, Lucia B Irazabal, Nazanin