Things used in this project
The Story behind the FallAssistant is the fact that the modern population is aging and at the same time we have more and more single households. Most of the accidents happen in the household and a quick help can be lifesaving. A sensor, which permanently monitors the position of the user with the ability to set up an emergency call after a potential fall, can be a solution for some of these problems.
The Hexiwear-Sensor-System provides all the necessary characteristics to solve this task. Within the 3D-printed enclosure, which model file is attached to this project, the Hexiwear can be clamped to the belt of the patient. The software is based on the NXP-Firmware, which can be downloaded from Github, and is programmed with the Kinetis-Design-Studio. The Android-App of the project is based on the Evothings-Examples which can be downloaded and customized with the Evothings-Studio. I used the Visual Studio 2015 for programming the App because adding PlugIns to your project is made there very easily.
I think that you have already paired your Hexiwear with your smartphone. If not, the original Hexiwear-App can do that job most easily for you. Then you have to install the HEXIWEAR_MK64.bin file from my linked repository on your Hexiwear-Device and add the fallAssistant-App to your Android-Smartphone. Please set off the Sensor-Tag-Mode of the Hexiwear, because the capturing rate in that mode is too low. Then you can start the Fall Assistant Mode on your Hexiwear.
After the Hexiwear is set in fallAssistant-mode, the accelerometer is detecting continuously the alignment of the patient and sends an alarm command over the BLE-Connection to a paired smartphone in the case of a fall. The fallAssistant-App running on the smartphone starts a call to a previously set contact, so the patient can request assistance over the phone speaker.
Usually the Hexiwear has to be worn upside-down, so the user is able to see the screen in the right direction. The bar-graph on the top part of the screen shows the alignment of the person. This graph gets green, when the user is in upright position. The fine blue line on the bottom part of the screen shows a life-view of the acceleration-vector of the sensor. You can see every direction changing on that dancing line. This line gets red if you exceed the configured acceleration limit. You can change the configuration values and limits of the fall detection in the Fall Config Menu and should check them before using the detector for the first time.
The detection algorithm is based on a few values. You can change the maximal capturing period with the poll delay setting. All occurred alarm limits within one capturing phase are added together to an alarm sum. If this sum exceeds a preconfigured limit, an alarm countdown will be started because the person might have fallen.
The countdown gives you time to cancel the alarm by touching the hand symbol-button or getting up again in the upright position. After the countdown is expired an alarm command ID with a time stamp and the captured threshold values will be transmitted over the Bluetooth connection to the paired smartphone. And the smartphone will start a phone call to the preset contact automatically.
All captured alarm data is stored in the internal Hexiwear EEPROM and can be viewed over the Fall Data Menu. Each alarm gets a time stamp, its configured limits and the measured acceleration values. The configuration of the fall detection can be improved by analyzing this results. It helps to avoid false alarms and get a better personal setting in the everyday life.
The fallAssistant-Android-App is very easy to use with its self-explaining text-to-speech mode. When you first start the app, it will ask you to set up an emergency contact, which you can be done with the yellow button. At this point you should give a read permission for your contacts to the app. After that the app tries to connect to a paired Hexiwear device. If a device is within range, you get a confirmation for the established connection. The connection itself can be handled manually with the red and green buttons.
Additional data can be seen in the app, like the last captured alarm ID, the alarm contact or the received sensor values. After the battery level undercuts the 10% value, you will hear an announcement which will inform you of a critical battery value. At this project stage the maximum activity time is less than five hours with the build in battery. It is advisable to use an additional battery pack to extend the runtime of the fallAssistant application. Furthermore, you could try to optimize the power save modes of the default Hexiwear firmware.
Here is a short video which captures the function of the Fall Assistant. Please consider, that the acceleration limits have to be achieved by the hips of the person which is wearing the sensor. So you can reach the limits by hand much easier.
Custom parts and enclosures
fallAssistant belt clip - 3D data
Did you replicate this project? Share it!I made one
Love this project? Think it could be improved? Tell us what you think!