intro

Introduction

‘Break My Fall’ is a wireless waistband that monitors whether a person falls over in an incapacitating manner. A mobile phone application receives information from the device and can alert a chosen person in a local or remote respect.

Full Product Brief PDF

further

Information Represented

A common form of injury sustained during infancy is a fall - for example, from a table, a bunk bed a ladder, or falling over when running. In the absence of a responsible adult, fall detectors could potentially allow response time to be critically shortened by an alert system. The device itself has been designed with children as the driving focal point, with particular design considerations for their movement and energy. A waistband attachment is quite seamless to children and would allow multiple communication of fall detection to one single responsible adult. The colour choices and playfulness of the project have been aesthetically engineered for use by children and people whose duty of care is children. A simple user interface utilizing existing mobile phones attempts to reduce the stress of monitoring children from a remote or local environment. The colour choice for the interface has this sensitivity in mind, and recognizes the importance when it comes to dealing with children.

inspiration

Brief

Families that hold a large number of children such as those in the rural areas of Philippines (Bagio City) require a high level of attention to each child with respect to care and protection. With increased responsibility and divided attention it is not always easy to focus your attention to each child’s wellbeing. This is evidently seen in cases whereby children are not in the same location as the parent or child minder, for example at school or a day care centre. Usually when the parents have to go to work, the babies are left with the adolescent children because child care services are not affordable.

Previous Research

There is research being performed into developing complex algorithms into analysing such falls, evidently seen in Garrett Brown’s 2005 paper “An Accelerometer Based Fall Detector: Development, Experimentation, and Analysis” His model monitors patterns of acceleration and sudden deceleration as if to mimic the stopping of movement when coming into contact with the floor.

(Brown,2005) Amplitude / Magnitude Algorithm for fall detection

concept

Concept

The bearer attaches the band to their waist and continues daily activities as normal. A person running the Break My Fall application on their mobile phone will connect to the waistband and be alerted if a fall is detected. The accelerometer will be able to decipher serious falls from non-serious falls. The accelerometer captures acceleration in the X, Y and Z axes and then rotation about each of these axes. This in turn will alert the nominated person via the option of SMS / Phone Call if remotely monitoring or sound an alarm if located within the vicinity. A logging service will allow the monitoring person to view time stamped occurrences of falls to view how their chosen bearer is progressing, for example in case of third party child care nannies. This log will be saved to a website for remote viewing by any personal computer and on the phone itself that is running the application. The person monitoring the bearer will be able to have control over the alert method and the run down timer to send an alert after a fall has occurred in the cases whereby the person is monitoring themselves, or a child holds a mobile phone on their persons if a monitoring adult is not present.

User Evaluation Testing Concept

User Evaluation Testing Plan PDF

A recorded observation will be undertaken to effectively measure the impact the fall detection device against its addition to the quality of life to a responsible adult in the care of a child under the age of 2. In order to do gauge the effectiveness of such a device, test subjects will be observed, recorded and studied to gain meaningful results and interpreted to determine the value of the alerting device.

The reason for the observation is to record participant’s interactions and activities and map out different tasks within a given environment. This in turn assesses the comprehensibility of the alerting phone device during the occupation of a foreground task. The foreground task provides a distraction, so the test subject can be affected by the alerting device and not solely focusing upon it, in other words waiting for their child to fall. This will allow effective measurement of fall detection, data trasnfer and reaction times. The key goal of this device is to shorten the time taken to discover and render assistance to a fallen child.

Dataflow

equipment

The hardware associated with this project will incorporate a Nokia N80 mobile phone capable of blue tooth communication with the “Sparkfun Six Degree of Freedom” accelerometer waistband. The accelerometer will be housed in a protective casing as to ensure no damage in the event of a fall. The component board that makes up the accelerometer requires 4 x AA batteries and the switch in the on position to pair with available blue tooth devices. The accelerometer casing was produced from acrylic plastic and epoxy resin which was measured to the dimensions to fit both the accelerometer and the battery powerpack. The bottom foundation is removable to change the batteries with a small side hole in order to switch the device on and off. The colour scheme is clear, thus the built in LED’s reflect off the casing indicating data transfer between the mobile phone and the device and that the power is flowing. The case itself is 11.5cm x 6.5cm x 4 cm in dimension, with a Velcro strap to attach to the child’s waist.

The Nokia N80 will run a Java language program called a MIDlet which is coded in an application called Mobile Processing. This MIDlet will provide a user interface for adjusting settings for the monitoring program as mentioned above and will display a history log over a chosen period of time. This log will also send a HTTP message to a server to make store on a server and allow online remote viewing. There are number of Java 2 Platform protocols that enable features with the Nokia N80. Specifically Break My Fall will utilize:

JSR82 – Enables J2ME applications to access Bluetooth networking services to the accelerometer device

JSR120 – Wireless messaging API 1.1 which includes support for SMS messaging, which is required in the event of a an emergency alert notification being sent to a contact phone number.

Coding

PHP + MYSQL database

The MYSQL database was created with one table “fallTable” and one column “fall” which held a data type of variable characters with a limit of 40 characters per entry. The PHP script essentially takes an input “message” from the mobile phone applet in the form of a string “fall detected @ timestamp date and time”. When users access the PHP script from a web browser, they are given a paralleled visual style seen throughout the mobile applet and website. This has been hardcoded around the MYSQL queries in order to hold a coherent visual representation between the database table and the mobile phone applet. The large green number represents number of falls and the small black print displays the time and date at which the fall occurred.

PHP SCRIPT

The java applet itself was coded within the Mobile Processing environment which allowed the creation of a compatible MIDlet which the phone could run and interact with Bluetooth communication, SMS, and internal phone features such as screen display, sound alert (MIDI) and vibration. The program was constructed with a number of states in which it could switch between in order to display different screens and interactive qualities. There was a separate screen for Settings, History, Connecting and Monitoring. The settings display allowed users to pick and choose which type of alerts would trigger in the event of a serious fall being detected. There is a checkbox for Phone Call / SMS / Vibration and Alert Sound, allowing a user to choose potentially 4 alert types. If a checkbox is selected, the user proceeds to an input screen where they have control over the phone number to call/SMS, the volume of the MIDI alert or duration of the vibration.

The algorithm to determine whether has occurred or not is primarily based upon (Brown, 2005) method for detecting a falls. The detection begins with the checking of any large sudden tilts or rotation around the waistband device, meaning the child has begun to fall. There are threshold values of minimum and maximum values for yaw, pitch and roll and if the value stream from the accelerometer exceeds these limits, a flag is raised. If a large acceleration occurs after this flag has been raised, this could mean the child has fallen quickly into that direction. The final means is to wait 500ms, store their current position variables and wait a further 500ms and check if their position has changed within a given threshold value of 30 + or 30- from their stored position. If they have moved quite considerably from this position, it is possible the child has gotten up and moved and is quite ok. On the other hand, if there is little (+-30) to no movement, this could mean an incapacitating fall has occurred. The 5 states of image sprites are directly embedded into these 3 stages of the fall algorithm, hence giving indication to the condition of the child. If a fall then is flagged, the preferred alerts are notified to the values set up in the settings interface.