SPEEDY: A Fall Detector in a Wrist Watch - PowerPoint PPT Presentation

1 / 26
About This Presentation
Title:

SPEEDY: A Fall Detector in a Wrist Watch

Description:

11/16/09. ETH Z rich. SPEEDY: A Fall Detector. in a Wrist Watch ... buzzer. Fall Detector for Elderly People TLD2003 Thomas Degen ETHZ. 11/16/09. 11 ... – PowerPoint PPT presentation

Number of Views:255
Avg rating:3.0/5.0
Slides: 27
Provided by: IFE3
Category:

less

Transcript and Presenter's Notes

Title: SPEEDY: A Fall Detector in a Wrist Watch


1
SPEEDYA Fall Detectorin a Wrist Watch
Technology Leadership Day 2003
  • Thomas Degen

2
Content
  • Motivation
  • Concept
  • Principles of Detection
  • Results
  • Conclusion
  • Demonstration
  • Outlook

3
Motivation
  • Health Monitoring is a very attractive
    application field for wearable computers
  • The market for elderly people is growing in both
    numbers and economic figures
  • Existing fall detectors are often not truly
    ubiquitous

4
Goals
  • SPEEDY
  • integrated in a wrist watch
  • easy to handle (one button)
  • innovative algorithm basedon acceleration
    sensing
  • simple and low power

5
Concept
  • Two sensors (ADXL202) with two axes each measure
    the static acceleration of the device(1g 9.81
    ms-2 in standstill)
  • There are six degrees of freedom for the arm
    movement
  • no knowledge of the devices bearing or
    orientation
  • We work with the Norm
  • independent of bearing

6
Principle of Detection
  • Monitor the wearers speed relative to the ground
  • how detect the speed relative to the ground when
    the orientation of the device is unknown?

-2
9.81ms
7
Principle of Detection
  • Monitor the wearers speed relative to the ground
  • how detect the speed relative to the ground when
    the orientation of the device is unknown?
  • The vertical speed is estimated by a rough
    approximation
  • only little computing power is needed

-2
9.81ms
8
Principle of Detection (2)
  • the approximation is fundamentally wrong for high
    accelerations towards the ground ( )

-2
9.81ms
9
Principle of Detection (2)
  • the approximation is fundamentally wrong for high
    accelerations towards the ground ( )
  • Integrating the approximation yields to different
    inaccuracies
  • norm and integral are not commutative
  • offset and noiseof the ADXL202

-2
9.81ms
10
Principle of Operation
  • Measuring theorientation
  • ADXL202
  • User Interface
  • input
  • 1 button
  • output
  • LED
  • buzzer

11
Principle of Operation
  • Measuring theorientation
  • ADXL202
  • User Interface
  • input
  • 1 button
  • output
  • LED
  • buzzer

12
Principle of Operation
  • Measuring theorientation
  • ADXL202
  • User Interface
  • input
  • 1 button
  • output
  • LED
  • buzzer

13
Visualization on the computer
  • wireless transmissionof the measurement data
  • calibrationof the sensors

emulation of the data analysis algorithms
14
Results
  • Recording duringa fast fall
  • detection of thefall is based on
  • 1) high velocity
  • towards the
  • ground
  • 2) impact
  • 3) inactivity

15
Results (2)
  • Recording during a strong handshake
  • no detection offall (phase 1)because of speed2

16
Results (3)
  • The success story
  • only 65 of possible dangerous falls
    weredetected
  • no false alarmduring a two-daytrial in real life

17
Conclusions
  • We developed a fall detector in the form of a
    wrist watch.
  • truly ubiquitous
  • no false alarm
  • high acceptance for the elderly
  • The device is best fitted for practical tests in
    social housing
  • data logging
  • wireless transmission
  • software analysis tool

18
Conclusions (2)
  • Yet, the device is not ready
  • detection rate only 65
  • power consumption too high (2.6 mA)

19
Demonstration
acceleration values
subject (m 33y)
20
Outlook
  • Using low-power acceleration sensors
  • Adding a pressure sensor device
  • Incorporating a small loop antenna

21
Outlook (2)
  • Antenna gain and orientation

22
Outlook (3)
  • New algorithm for velocity towards the ground
  • no approximation
  • low computational power

23
SPEEDY A Fall Detector in a Wrist Watch
Thank you for your attention.
24
Hardwareplatform
  • MSP430F149 Controller
  • 2 Accelerometer ADXL202 (2g), 3 Raumachsen
  • Serielle Schnittstelle, Taste, Beeper, Led
  • Batteriebetrieb ca. 2 Wochen
  • Daten-Logging 40s (Flash)
  • Mit Wireless-Modul

25
Demonstration
Nur Fall (Speed1) und Aufschlagsdetektion
  • Ruhezustand Led blinkt langsam (1Hz)
  • Erschütterung
  • Fall Led blinkt schnell (5Hz)
  • Fall und Erschütterung - Sirene heult

26
Demonstration
Nur Fall (Speed1) und Aufschlagsdetektion
  • Vorgeschichte unnötige Todesfälle alter
    alleinlebender Personen (Oberschenkhalsbruch,
    Fettembolie)
  • GSM, (GPS) für Positionsangabe, Rettungseinsatz
Write a Comment
User Comments (0)
About PowerShow.com