Our uDirect technology can accurately estimate the facing direction of a mobile phone user, independent of device position and orientation and without any user intervention.
Knowing which way a user is facing can provide valuable information for the provision of mobile services and applications. The proposed technology utilises inertial sensors that are readily available in mobile consumer devices (e.g. smart phones). While providing high accuracy, the solution is able to cope with arbitrary wearing positions and orientations of the mobile consumer device, making it suitable for use in every-day life situations. This is achieved by an estimation of the user orientation with respect to the reference frame of both sensing module, and earth coordinate system.
Udirect: accurate and reliable estimation of the facing direction of the mobile phone users
1. 1
uDirect: A Novel Approach for
Pervasive Observation of User
Direction with Mobile Phones
Dr. Amir Hoseini-Tabatabaei
Dr.Alex Gluhak
Prof. Rahim Tafazolli
Centre for Communication and Systems Research
– University of Surrey
2. Outline
Introduction
Application of pervasive direction estimation
Pervasive direction estimation techniques
uDirect
Approach
Requirements
Algorithm design
Evaluation and Results
Limitation and Future work
Conclusion
2[1]
3. Application of Pervasive
Direction Estimation
Direction estimation is an essential part of a verity of
applications:
– Dead reckoning
– Human computer interaction(HCI)
– Social signal processing
– And many others
[2]
4. Pervasive direction estimation
techniques
Sensing Approach Typical sensors Shortcomings Advantages
Ambient sensors
UWB, Wi-Fi,
Bluetooth,
Camera,
Dependency to
infrastructure.
(localized
applications)
No power and
computation
limitation. No
limitation on time
Wearable sensors
Camera,
Accelerometer,
Magnetometer and
Gyro , GPS, IR ,
…
Short time data
collection
(Intrusiveness),
limited
computation and
energy resource.
No limitation on
location .
Mobile phone –
based
Accelerometer ,
Magnetometer ,
Gyro, GSM,GPS
limited
computation and
energy resource
and phone context
problems.
No limitation on
location and time
[3]
5. Mobile phone based
techniques and limitations
* Less accurate and more computational expensive than PCA of accelerometer[3]
5
Techniques Limitations
1. Principal Component Analysis (PCA)
based approaches • Limited to trousers pocket.
• Requires segments of unidirectional
movements with appropriate amount of
samples.
• Susceptible to outliers (not reliable)Gyroscope
measurements[5]*
Accelerometer
measurements[1-3]
2. Heading direction with absolute
positioning form Global Positioning
Systems (GPS)
• Susceptible to shadowing
• Limited positioning accuracy (e.g. 8
meter for mobile devices)
Assumption: People normally walk forward
[4]
7. uDirect : Requirements
Requirements
1- Providing a pervasive observation of the user facing direction on
mobile phones
2- Addressing the shortcomings of current approaches. (GPS and PCA
base models)
3- Addressing the related phone context problems
7[6]
8. uDirect: Approach
Estimating user orientation in a global coordinate:
8
Calibration
• Mobile –Earth
Local Direction Estimation
• User-Mobile
Global Direction estimation
• User-Earth
X
Y
Z
N
E
-G
[7]
9. Calibration
uDirect : algorithm design
Algorithm: performing estimations in two step
9
Utilizing the acceleration
pattern of the body segment
(corresponding to device
position) for identifying
proper moment s in
measured data in which
user orientation relative to
mobile can be estimated.
[8]
10. Calibration
uDirect : algorithm design
Algorithm: performing estimations in two step
10[8]
Utilizing measurements
from Accelerometer and
Magnetometer to estimate
the relative orientation of
phone and earth coordinates
11. 1. Calibration
Orientation: calibrating sensor readings with respect to the
reference coordination (Earth)
– Detecting earth coordinates
Gravity: Accelerometer
North : Magnetometer
– Calculating transformation components
Computational efficient form by using
mathematics of Hilbert’s space and quaternion
11
Dx
Dy
Dy
-G
N
E
H
Inclination
Declination
[9]
12. 2. Direction Estimation
We need an estimation of user coordination.
– We only have the vertical (V) direction form calibration.
How to find the F and S ?
First assuming the user coordinate is known:
I. What mobile phone measures during forward walking .
II. Transfer the measurements back to user coordinate .
III. Focus on behaviour of horizontal components during walking locomotion.
12[10]
13. 2.1. Direction Estimation
Assuming the mobile is in user’s trousers pocket.
1- Modelling accelerometer measurements caused by thigh movement
during walking locomotion.
Acceleration on mobile coordinate
yrzAycosxAysin
zrzAzsinysinyAcosxAycoszsin
zryrzA)t(ysin)t(zcosyA)t(zsinxA)t(ycos)t(zcos
22
oa )t(A
[11]
Rotation Quaternion : R(t)
Rotational Acceleration: Ar(t)
Translational Acceleration: Ah(t)
Aoa = Ar(t) + R(t)(Ah(t)+G)R(t)*
15. 2.3.Direction Estimations
Femur transverse rotation[5]
Polynomial fitted curve
Transverserotation
deviation(D)
Percentage of walking cycle
Reconstructed
Acceleration(m/s^2)
z
Swing Phase Stance Phase
Heel Strike Toe off
[13]
3- Finding the proper moment for estimation
Heel Strike and toe of moments can be detected as local and global
minima of tight vertical acceleration pattern[6].
18. Results. II
Algorithm performance
Performance in Comparison
with GPS approach
Baseline
PCA[1]
uDirect (Average per section)
Techniques
Mean error (Degree)
Section1 Section 2 Section 3 Section 4
Model from
[1]
26.7 37.6 10.5 37.0
uDirect(averag
ed per section)
18.9 41.7 37.3 35.4
Technique
Mean error
(Degree)
Standard
deviation
Model from [1] +7.999 +1.253
uDirect(averaged
per section)
+0.162 -0.603
uDirect (averaged
per step)
-10.5 -11.8
DeviationfromNorth(D)
Steps
19. Limitation and future works
• Similar to conventional PCA based approach the current
model is limited to trousers pocket
Extending the approach to other main positions[7] : shoulder
bags, chest pocket and belt – enhancement positions
• The estimations degrades in shorter sections
To adaptively select the estimation model
To add the magnetic field based-tracking for reducing
power consumption.
[16]
20. Conclusion
Developing and evaluating the uDirect as a direction estimation
techniques for mobile phones.
The model is based on physiological characteristics of human walking
locomotion.
Evaluations of the algorithm with a simple proof of concept
implementation confirmed the assumptions of our analytical modeling
Orientation independent approach
Dose not face with GPS approach constrains (shadowing and
minimum distance)
Direction estimations in contrast to PCA do not require an additional
segmentation.
Independent estimation at each step makes it prone to error
accumulation.
uDirect is shown to be more accurate and reliable than conventional
GPS and PCA based models for paths longer than 2 steps
20[17]
22. References
[1] K .Kunze, P. Lukowicz, K. Partridge, and B. Begole, "Which way am i facing:
inferring horizontal device orientation from an accelerometer signal," in Wearable
Computers, 2009. ISWC '09. International Symposium on, Linz, 2009, pp. 149-150.
[2] M. Kourogi and T. Kuratta, "A wearable augmented reality system with personal
positioning based on walking locomotion analysis," in Proceedings of the 2nd
IEEE/ACM International Symposium on Mixed and Augmented
Reality, Tokyo, 2003, p. 342.
[3] U.STEINHOFF, B.SCHIELE,2010. Dead Reckoning from the Pocket - An
Experimental Study. In Eighth Annual IEEE International Conference on Pervasive
Computing and Communications (PerCom2010), 2010.
[4] U. Blanke and B. Schiele, “Sensing location in the pocket,” in Adjunct Poster
Proceedings UbiComp’08, 2008.
[19]
23. References
[5] A.S. Levens, V.T. Inman, and J.A. Blosser, "Transverse rotation of the segments of
the of the lower extrimity in locomotion," The journal of bone and joint surgery, vol.
30, pp. 859-872, 1948.
[6] K Aminian, K. Rezakhanlou, E.D. Andres, and C. Fritsch, "Temporal feature
estimation during walking using minitaure accelerometer: an analysis of gait
improvement after hip arthoplasty," Journal of Medical & Biological Engineering
& Computing, vol. 37, no. 6, pp. 686-691, 1999.
[7] F. Ichikawa, J. Chipchase, and R. Grignani, "Where's the phone? A study of mobile
phone location in public spaces," in International Conference on Mobile
Technology, Applications and Systems, 2005 2nd, Guangzhou, 2005, pp. 1-8.
[20]
Notas del editor
Talk about
order to adequately render human to computer interaction
This approach is very close to wearable approachSuffer from similar However is not considered intrusive and is not faced with such limitations in data collection time.
The accuracy of the
What ever we sample form user in different time would be in different positions and orientations
Get your attention
A walking cycle consist of two phases : stance phase and swing phase.
General form
Idea was to have horizontalEquationLooked through the empirical measurments of rotation of related body segment