An ETextile System for Motion Analysis

1
An E-Textile System for Motion Analysis

• Mark Jones, Thurmon Lockhart, and Thomas Martin
• Virginia Tech

2
Virginia Tech e-Textiles Group

• Design of an e-textile computer architecture
• Networking
• Fault tolerance
• Power aware
• Programming model
• Design through simulation
• Emulation/Simulation environment
• Across population
• Development of application prototypes

3
Application Motivation

• Falls are one of the leading causes of death
  among the elderly in the U.S.
• Only 50 of those hospitalized with fall-related
  injuries survive their next year
• Hip pads for at-risk patients are bulky and
  inconvenient, leading to low compliance rates
• E-textiles have been shown to have significant
  potential in the health care field
• Our goal is to develop an e-textile solution that
  will achieve high compliance rates

4
Gait Analysis

• Gait analysis can identify patients at risk for
  falling as well as several pathological
  conditions
• Currently performed in dedicated laboratories at
  high expense
• Somewhat artificial
• Time consuming

Virginia Tech Locomotion Laboratory
5
Measures in Gait Analysis

• Raw Data
• Position (x,y,z) of the body
• Force of foot-to-ground
• Gait measures
• Stride length
• Required coefficient of friction
• Transition of center of mass
• Width of gait

6
E-Textile for Gait Analysis

• We are building an e-textile system with the
  following features
• Pants augmented with sensors
• Footwear with two force sensors
• Hip airbag for the pants
• Remote communication device
• Advantages no time for setup, can be used in
  home environment, mitigates fall impact, users
  more likely to be compliant, more natural
  measurements
• The design issues identified are discussed in the
  following slides

7
How to Obtain Gait Measures

• The sensors under consideration (accelerometers,
  force sensors, angular velocity sensors,
  gyroscopes) do not directly sense any of the gait
  measures
• We propose that a combination of sensors,
  combined with computation, can determine these
  gait measures
• Design Issue What is the set of sensors that
  will provide these measures at an acceptable
  accuracy level?

8
Designing for the Masses

• The proposed system must work across a range of
  sizes and gait types
• A single weave design for the bolts of cloth
• Standard garment sizes constructed from that bolt
  of cloth
• Sensors will be in slightly different positions
  on each user due to motion and size differences
• Range of sensor readings will vary across users
• Design Issue It is not practical to assume that
  we can construct and test prototypes for a range
  of users repeatedly while exploring the design
  space

9
Application Functionality

• What is required to provide informative data?
• In the gait analysis laboratory, the system is
  only triggered for a brief period of time as the
  user is in the correct location and walking
• In a doctors office, we need to record and
  analyze data only during a specified period
• Avoid time-consuming data searching
• In a home setting we need more automation
• Must identify when a user is walking, then
  trigger recording
• Must identify when a user is falling, then
  trigger air bag

10
Exploring the Design Space Through Simulation
11
Current Status of Garment

• We have fabricated a pair of pants for motion
  classification
• Designed through simulation
• Trained neural network across a range of virtual
  users
• Tested the pants successfully on the first real
  wearer
• Worked with NN trained via virtual users
• Features of our architecture
• All digital communication
• Fault tolerant
• Power aware operation
• On-garment computation and decision making

12
Computing Gait MeasuresStride Length Example

• Accelerometer on each ankle
• Identify begin/end of stride in the data (force
  sensors will be used for more accuracy later)
• Integrate the acceleration value twice to find
  the distance traveled by the ankle
• Gait analysis studies provide us with the data to
  determine what is significant error
• For example, we can use the mean heel velocity in
  two subject groups as well as the standard
  deviation of heel velocity

13
Conclusions and Future Directions

• E-textiles hold great promise in improving the
  usability and acceptance of home health care
  devices
• Cross-disciplinary teams are essential
• Design for cost-effective fabrication may allow
  for wider spread adoption
• Simulation can be very effective in the design
  process
• Common architecture can speed design and
  deployment
• Gait analysis is an area where early impact of
  e-textiles is possible
• Evaluation and deployment plan is essential