Transcript of ToBITas Case Study, Presentation for UCAMI 2014 conference
- 1. Evaluation of a Context-Aware Applicationfor Mobile Robot
Control Mediatedby Physiological Data: The ToBITas Case StudyBorja
Gamecho1, Jos Guerreiro2, Ana Priscila Alves2, Andr Loureno2,Hugo
Plcido da Silva2, Luis Gardeazabal1, Julio Abascal1, Ana L. N.
Fred21 University of the Basque Country (UPV/EHU)Egokituz:
Laboratory of HCI for Special Needs2 Instituto Superior Tcnico
University of Lisbon (IST-UL)PIA: Pattern Image and Analisys
Group
- 2. 2/27Outline Motivation Research goals System description:
What is ToBITas ? Evaluation: Usability study Conclusion
- 3. Proliferation of Devices with Embedded Sensors3/27
Smartphones/Tablets A box with sensors Wearable Devices User
activity and user emotional state recognitionWearables expands the
sensing opportunities forsmartphone applications
- 4. Take advantage of the growing ecosystem ofsensor devices
Make it easier for developers to combine sensorsignals and new
context discovery Support the generation of context-informationfrom
the sensor data Enhance the interaction between users andmobile
applications4/27Our Proposal
- 5. Framework for Mobile Context Awareness5/27
- 6. Example Application Using the
Framework6/27MobileBITFrameworkBITALINO ROBOT CONTROL
- 7. Research GoalsTesting the satisfaction and adaptation of
users7/27to physiologically-enhanced sensors inputmethodsCreating
an application based on low-cost sensorplatforms to extend the
smartphone sensingcapabilities using a Context-Aware
approachTesting the feasibility of the MobileBIT frameworkto create
Context-Aware applications
- 8. Context-Aware application to control a mobilerobotic
platform using physiological sensors.8/27System Description
- 9. 9/27Devices for ToBITas
- 10. 10/27Devices for ToBITas Wirelessly operated mobile
robotplatform Programmed using Arduino Bluetooth interface Receives
commands from theSmartphone to move and use the clawBotn RollMobile
RoboticPlatformRobot commands
- 11. 11/27Devices for ToBITasBITalino SensorPlatformOptimized
for real-time data streamingSampling Rates: 1, 10, 100, or 1000
HzBluetooth classic connectivitySensors Available:- EMG :
Electromyography- ECG : Electrocardiography- EDA : ElectroDermal
Activity- ACC : Accelerometer- LUX : PhotodiodeElectromyography -
EMG Technique to measure the electricalactivity produced by the
muscles 2 channels for TobitasAccelerometer ACC Device to measure
g-force 1 axis (Z) for ToBITas
- 12. 12/27Devices for ToBITasLG Optimus F5 Dual-Core 1.2 Ghz 1
GB RAM Bluetooth Communications Android 4.1.2SmartphoneAppFor
ToBITas Application: Implements MobileBIT framework Manages
Bluetooth connection withBITalino and Botn Roll Process data from
sensors andtransforms into context-information
- 13. 13/27Devices for ToBITasHuman Body:Right Arm
- 14. Signal User Movement Context-Information Robot
Command14/27EMG_1 The user folds his arm
Action_detected:Right_arm_foldedMove ForwardEMG_2 The user closes
hishandsAction_detected:Hand_ClosedOpen/Close the ClawACC Tilt the
forearmPosition_detected:Forearm_upMove
RightPosition_detected:Forearm_downMove
LeftPosition_detected:Forearm_sideDont MoveRemote Control
- 15. Data Processing Signals acquired at 100Hz Every context
command is detected based on blocks of40 samples (with no
overlapping) EMG signal data processing:15/27 ACC signal data
processing: Low-pass filter implemented using a moving
averageapproache1, e2,
- 16. 16/27EMG SignalData over the threshold = Intentional
Movement
- 17. 17/27ACC SignalLEFT LEFT LEFTRIGHTCENTER
- 18. 18/27MobileBIT FrameworkMobileBIT Framework
- 19. 19/27Evaluation Usability Test: Quantitative and
Qualitative Evaluate the adaptation of the user
tophysiologically-enhanced sensor inputmethods 13 Participants (4
females) in 3 groups: Novices (x7): No prior experience Experienced
(x4) : Have used EMG/ACC control before Experts (x2) : Involved in
the develop and test of Tobitas
- 20. 20/27Research Questions Are the users able to control our
system ? Complete a task in a reasonable time Measure learning
effect curve Do the users feel comfortable with this kind ofcontrol
? Complete the SUS test [Brooke 1996] Scores above 68 are
considered above average[BROOKE 1996] - Brooke, J. (1996). "SUS: a
"quick and dirty" usability scale".In P. W. Jordan, B. Thomas, B.
A. Weerdmeester, & A. L. McClelland. UsabilityEvaluation in
Industry. London: Taylor and Francis.
- 21. 21/27Evaluation: Task DescriptionRepeat thetask
threetimes
- 22. 22/27Experimental Setup
- 23. Experimental Results: QuantitativeTable 2. Summary of the
task performance results for each group of participants(measured in
seconds).23/27NovicesExperiencedExpertsAll160140120100806040200Attemp
1 Attemp 2 Attemp 3Novices (A)Experienced (B)Experts (C)All
(A+B+C)Seconds
- 24. 24/27Experimental Results: QuantitativeVisualization of all
the trials for each groupNovices (A) Experienced (B) Experts
(C)Seconds
- 25. 25/27Experimental Results: SUS Scores Group A and Group B
(11 participants) Average score: 73.86 A SUS score above a 68 would
be considered above average and anything below 68 isbelow average.
(http://www.measuringu.com/sus.php) 7 participants over 70 (63.6%)
5 from Group A and 2 from Group B 3 participants from 60 to 70
(27.3%) 1 from Group A and 2 from Group B 1 participant lower than
60 (9.1%) From group A: Lowest thresholds in the
calibrationphase
- 26. 26/27ConclusionAdaptation and satisfaction: Learning effect
can be noticed: Adaptation After third run values are similar for
group A and B A and B finished in reasonable time The time was 44%
longer than the experts group The system is perceived as usable SUS
score has been over 68 in averageSmartphone sensors have been
extended usingthe BITalino sensor platformToBITas is a functional
and usable Context-Aware application
- 27. 27/27Future Work Exploratory study shows promising results
More users are needed to reinforce our claims Flaws detected Three
signals activate simultaneously: Arm folded Hand closed Arm
position right Improved signal processing algorithms arerequired
for fine tuning
- 28. Experimental results and this presentation isavailable at
:http://borjagamecho.info/tobitasEvaluation of a Context-Aware
Applicationfor Mobile Robot Control Mediatedby Physiological Data:
The ToBITas Case StudyBorja Gamecho1, Jos Guerreiro2, Ana Priscila
Alves2, Andr Loureno2,Hugo Plcido da Silva2, Luis Gardeazabal1,
Julio Abascal1, Ana L. N. Fred21 University of the Basque Country
(UPV/EHU)Egokituz: Laboratory of HCI for Special Needs2 Instituto
Superior Tcnico University of Lisbon (IST-UL)PIA: Pattern Image and
Analisys Group