This is an Instructable for a monitoring and alerting device for someone such as an elderly person who may be on their own at times during the day or night and who may be at risk of falling or may require some assistance during that time. The concept for the device is based on an Intel Edison Arduino board and was built and demonstrated during an IoT Challenge week which was kindly hosted by Intel in their UK Ignition lab in Swindon.
The initial concept demonstrator uses a 3 axis accelerometer to detect a possible fall. On detection of a possible fall the device sounds a warning buzzer. If the the warning was triggered accidentally, the wearer can cancel the buzzer using one of two push buttons. If the warning buzzer is not cancelled with a few seconds a fall alert is automatically triggered and a message is sent to a Cloud service via the WiFi network. The device has a second button which can be used to trigger a panic or assistance alert. When pressed, the buzzer will sound for a few seconds giving the person time to cancel the buzzer in case of false alarm. If the buzzer is not cancelled, an assistance alert message is triggered and sent to the Cloud service. The Cloud service can then be configured to send push email notifications and SMS messages according to defined rules such that neighbours, carers or relatives may be notified according to the alert condition. Emails and SMS messages can be sent directly from the device but using a cloud service allows easy modification of alert notification rules for cases when people are on vacation or are not able to respond.
When an alert has been triggered, the device buzzer stops its pulsed tone and reverts to a quieter pulse of sound every few seconds to notify the person that the messages have been sent. This sound can be cancelled by recipients of the notifications by logging onto the Cloud service to let the person know that assistance is on its way. The sounder can also be cancelled by holding the Cancel button down several seconds.
For the Intel IoT Challenge week, a small watch like housing was built to demonstrate a future possible concept and the longer term aim is to be able develop the concept into a wearable or pocket-able device with additional sensors such as temperature and pulse rate as well as visual indication such as a small LCD or OLED screen.
Step 1: List of Parts
The parts used for the initial concept demonstrator are as follows: -
1 x Intel Edison Module and Arduino Breakout Board
1 x DC Power Supply
2 x Grove Button
1 x Grove Buzzer
1 x Grove 3-Axis Digital Accelerometer (MMA7660 +/-1.5g)
Note that other buttons and buzzers can be used. Other accelerometers can be used but the code will need to be changed to suit.
Additional parts for the wearable in this particular project are:-
1 x 3D printed housing
2 x Tactile switches (eg Maplin type 105A)
1 x Nylon webbing strap buckle (16 mm)
30cm of Nylon webbing strap (16 mm)
Wire and glue as required.
Other Resources Used
Cloud service for Intel IoT devices - Enable IoT (Free signup)
SMS Messaging Service - Optional, to send SMS via HTTP or email trigger (subscription required) BulkSMS used for the demonstrator
Step 2: Setting Up
Step 3: System Configuration
The sensor setup for the Edison is relatively simple requiring just three GPIO pins for the buzzer and two switch buttons, and one I2C port for the accelerometer. A PWM capable pin is required for the buzzer and is connected to D3 on the Arduino breakout board. D6 and D7 are used for the two switch buttons.
The watch-like housing had two tactile push buttons glued to the back plate which sat below the Alert (Red) and Cancel (Yellow) buttons on the housing. These were piggy backed onto the two Grove button boards plugged into the Arduino breakout using three wires (VCC, SW1 and SW2).
Ideally the accelerometer IC would have been mounted in the same housing but it is a very small BGA/QFN package so the cable to the module was extended instead so that it could be attached close to the housing.
Similarly the buzzer was de-soldered for the PCB and extended with flying leads.
Step 4: Fall Detection Algorithm
Accurate detection of a fall is quite a complex and there are several articles available on the web such as "Detecting Human Falls with a 3-Axis Digital Accelerometer", by Ning Jia of Ananlog Devices
For the purposes of the proof of concept demonstrator a crude fall detection method was used which was based the free fall element of the article described above.
Step 5: Code
Step 6: Testing and Operation
Here is a quick video showing the monitor being tripped and reset. The screen capture shows the Enable IoT website messages posted by the monitor.
Step 7: Possible Future Developments
The Intel Edison and sensors used for the concept demonstrator could easily be packaged into a wrist wearable device. However, the battery capacity to power for the Edison running continuous accelerometer algorithms for a reasonable duration is likely to be impractical for a wrist wearable device. The monitor could still be made pocket-able and could incorporate other features such as an easily readable screen, a vibration motor and two way audio communications, for example.
As an alternative, some digital accelerometers can be configured to send an interrupt if preset g levels are exceeded. Although the fall detection algorithm would not be so comprehensive, this approach along with the switch buttons configured as GPIO interrupts would enable the Edison to be run in a lower power state thus requiring a smaller battery.
The fall monitor concept was constructed and demonstrated during a one week IoT challenge so there is definitely scope to take this further and enhance it's capability. I am interested in building a more practical working prototype but think that the small scale surface mount devices and fine pitch leads would be too much of a challenge right now.
There are a growing number of Bluetooth LE fitness trackers that have built in pulse and g sensors with some manufacturers encouraging developers. This approach in conjunction with an Edison based base-station to replace the Smartphone would seem to provide a much more flexible system.
8th Aug 2015 -Added video clip recorded at Intel Ignition labs
-- Work In Progress --