Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.
-
Upload
prudence-james -
Category
Documents
-
view
221 -
download
3
Transcript of Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.
![Page 1: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/1.jpg)
Mitja LuštrekJožef Stefan Institute
Department of Intelligent Systems
![Page 2: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/2.jpg)
Environment should be◦ Intelligent◦ Require no special skills of the user◦ Require minimal interaction from the user
The technology should disappear Its advantages should remain
Defined by objectives, not methods Interdisciplinary
![Page 3: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/3.jpg)
On the go:◦ Wearable sensors◦ Smart phone applications
At home:◦ Sensors◦ Computer controlled appliances◦ Home automation
Living labs (Philips...)
![Page 4: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/4.jpg)
Pupulation is aging – over 65 in Europe:◦ 17.9 % in 2007◦ 53.5 % in 2060
Not enough young people to care for the old Technology must step in
◦ Assistance with activities of daily living (ADL)◦ Detection of health problems
![Page 5: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/5.jpg)
Equip elderly with radio tags
Sensors determine tag coordinates:◦ Installed in the
appartments◦ Included in tags and
portable device outdoors Detect falls and other
health problems
Portable
device
Body tags
Sensors in the
appartment
![Page 6: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/6.jpg)
Equip elderly with radio tags
Sensors determine tag coordinates:◦ Installed in the
appartments◦ Included in tags and
portable device outdoors Detect falls and other
health problems
Intelligence
![Page 7: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/7.jpg)
Radio tags and sensors to be developed in the project◦ Distance to tag – time needed for signal to travel
from tag to sensor◦ Direction of tag – angle of arrival of the signal
Expected standard deviation of noise:◦ ~5 cm when stationary (Ubisense × 1)◦ ~10 cm when moving (Ubisense × 2)
![Page 8: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/8.jpg)
6 infrared cameras 12 reflective markers on the
body Multiple cameras see a marker
⇒ location can be computed
Standard deviation of noise:◦ ~1 mm
Add more noise to simulate radio hardware
![Page 9: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/9.jpg)
815 recordings:◦ Walking◦ Sitting◦ Lying◦ Falling – 11 types◦ Lying down◦ Sitting down◦ Health problems:
Limping Hemiplegia (stroke) Parkinson’s disease Dizziness Epilepsy
Six
basic
activ
ities
![Page 10: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/10.jpg)
Input: sequence of snapshots of tags (each consisting of coordinates of all tags)
Attributes
Output: posture/activity (walking, lying...)
Class
![Page 11: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/11.jpg)
Manually segment and label recordings Compute attributes for each snapshot Concatenate to create attribute vectors
![Page 12: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/12.jpg)
Z coordinates of tags Absolute, z velocities of tags Absolute, z distances between tags
Attributes – Attributes – anglesangles
![Page 13: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/13.jpg)
All coordinates of tags Velocities of tags
(absolute, direction)
One coordinate systemper snapshot
One coordinate systemper 1-second interval
Two options
Two more options: each coordinate system can use reference z axis
![Page 14: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/14.jpg)
![Page 15: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/15.jpg)
Attributes: reference coordinate system Machine learning algorithms:
◦ SVM◦ Random forest◦ Bagging◦ Adaboost M1 boosting◦ 3-nearest neighbor
Winner:◦ Reference coordinate system + angles◦ SVM
![Page 16: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/16.jpg)
Sitting down, no noise
Falling, Ubisense × 1 noise
![Page 17: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/17.jpg)
Tag placement◦ More tags ⇒ better performance◦ More tags ⇒ worse user acceptance
Noise level◦ We are only estimating noise of the radio
hardware
![Page 18: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/18.jpg)
12 11 10 9 8 7
6 5 4 3 2 1
LR
![Page 19: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/19.jpg)
![Page 20: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/20.jpg)
We can recognize walking Can we recognize abnormal walking?
Gait (way of walking) important to physicians
Used to recognize health problems in clinical setting
![Page 21: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/21.jpg)
Support (foot on the ground), swing (foot off the ground) and step (support + swing) times
Double support time (both feet on the ground) Step length and width Maximal distance of the foot from the ground Ankle, knee and hip angles upon touching the ground Knee angle when the ankle of the leg on the ground is
directly below the hip and knee angle of the opposite leg at that time
Minimal and maximal knee and hip angles, the angle of the torso with respect to the ground, and the range for each
Hip and shoulder sway (the difference between the extreme left and right deviation from the line of walking)
![Page 22: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/22.jpg)
X, y coordinates of ankles L: lowest distance travelled (standing still) H: highest distance travelled (moving)
![Page 23: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/23.jpg)
Normal:◦ Completely normal◦ With a burden
Abnormal:◦ Limping◦ Hemiplegia (stroke)◦ Parkinson’s disease◦ Dizziness
![Page 24: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/24.jpg)
![Page 25: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/25.jpg)
![Page 26: Mitja Luštrek Jožef Stefan Institute Department of Intelligent Systems.](https://reader036.fdocuments.in/reader036/viewer/2022062717/56649e455503460f94b3a83f/html5/thumbnails/26.jpg)
In-depth analysis of activities other than walking
Attributes other than walking signature
Macroscopic movement (about the appartment)