We present a new approach, called PseudoSensor, which emulates otherwise inaccessible input modality by repurposing the existing sensors in mobile devices. In order to demonstrate our concept, eight applications are introduced in this thesis. NailSense and CamPress are interaction techniques that repurpose a camera sensor to emulate a pressure sensor. MicPen and PseudoButton are software solutions that repurpose a microphone to determine pressures exerted by users. ForceTouch and VibPress are software techniques that repurpose an accelerometer sensor to determine the level of pressure users apply to a mobile device. MagGetz and MagPen are software techniques that repurpose a magnetometer sensor to determine the level of pressure users exert on various types of input widgets. These applications explore the possibilities of emulating an input modality, focusing especially on emulating pressure sensors, which current mobile devices do not possess. By testing a series of applications and showing the results of user studies, we demonstrate the feasibility of our approach and the fact that the empirical evidence indicates high levels of performances. Finally, we distill our concept into a unified guideline and formalize it using some basic elements.
4. •Current mobile devices are equipped with a variety of sensors, offering numerous input channels for expressive interaction techniques.
•Researchers have leveraged on these capabilities for creating new input interaction modalities to enhance the user experience and explored the design space to utilize them.
BACKGROUND
(http://www.noahlab.com.hk/labvision/research_hci)
(http://www.qualcomm.com/news/snapdragon/2014/04/24/behind- sixth-sense-smartphones-snapdragon-processor-sensor-engine)
5. Issue 1. Some sensors are impractical for small devices (e.g., wristwatches, glasses, rings) due to the limited computing power and small interaction area available.
Issue 2. Some sensors (e.g., pressure, touchscreens, humidity) are seldom available on mobile devices today; instead they have to be attached to the device as bulkyaccessories, which ordinary users have seldom available (Maragos, Potamianos, & Gros, 2008).
Miyakiet al., 2009, Esslet al., 2009, Heoet al., 2011, Wilson et al., 2013
BACKGROUND : ISSUES
Images from Google, Fin ring, Pebble, and Samsung website (Nov. 20. 2014)
(a)
(b)
6. Issue 3.Augmentation of input hardware causes additional production costs for device vendors and additional maintenance cost for users.
http://www.ifixit.com (Nov. 20. 2014)
BACKGROUND : ISSUES
7. “What if we emulateunavailable sensors through available resources on mobile device?”
“What if we generate unavailable sensors through a software approach?”
This study starts from the question,
BACKGROUND : RESEARCH QUESTION
8. Schenkman, B. N., and Nilsson, M. E. (2010). "Human echolocation: blind and sighted persons’ ability to detect sounds recorded in the presence of a reflecting object," Perception (39:4), p 483.
Hint from Human : Human echolocation
A blind person recognizes surrounding objects by detecting sound recorded in the presence of reflecting objects.
BACKGROUND : RESEARCH QUESTION
9. GOAL 1.
To present a concept that repurposes input resourcesof the mobile device.
GOAL 2.
To empirically provethis method through various instance applications.
GOAL 3.
To build a unified set of guidelinesthat a broad range of HCI could utilize.
RESEARCH GOAL
12. PRESSURE-BASED INPUT METHOD FOR MOBILE DEVICES
1.It adds a degree of freedom to the touch locations (Boring, Ledo, Chen, Marquardt, Tang, & Greenberg, 2012)and so it can free the user from spatial restrictions and repetitive movement(Miyaki, & Rekimoto, 2009).
2.Pressure input allows relatively stable and accurate interactions when user is in mobile context(Wilson, Brewster, Halvey, Crossan, & Stewart, 2011).
3.Pressure input supports in-pocket operation. For instance, user can interact with the device when it is in the pocket or bag.
4.It can be used to alleviate occlusion problems and can also provide ways for rich contextual selections(Ramos, Boulos, & Balakrishnan, 2004).
Several advantages of pressure input for mobile device
Previous work on pressure input techniques for mobile devices either introduced specialized sensing hardwareor relied on softwareto estimate the pressure from sensors commonly available on mobile devices
13. HARDWARE-AUGMENTATION APPROACH TO ENABLING PRESSURE INPUT
GraspZoom
(Miyakiet al., 2009)
Pressure-based Text Entry
(Brewster et al., 2009)
Squeezing the Sandwich
(Essl, 2009)
Pressure-based Menu Selection
(Wilson et al., 2010)
ForceGestures
(Heo et al., 2011)
Indirect Shear Force
(Heoet al, 2013)
Multi-digit Pressure Input
(Wilson et al., 2012)
ForceDrag
(Heo et al., 2012)
15. SOFTWARE APPROACH TO ENABLING PRESSURE INPUT
A tangible controller(Kato et al., 2009)
Vision-based Force Sensor(Sato et al., 2012)
GripSense
(Goelet al., 2012)
The Fat Thumb
(Boring et al., 2012)
c
c
t
g
Muscle Tremors(Strachan et al., 2004)
Use the Force or something
(Esslet al., 2010)
ForceTap
(Heo et al., 2011)
Expressive typing
(Iwasaki et al., 2009)
c
t
a
g
camera
touchscreen
accelerometer
gyroscope
As a response to the limitation of hardware augmentation approach, some authors have attempted to estimate input pressure with software, using readings from sensors available on most mobile devices.
a
a
a
a
16. Sonicstrument
(Lee et al., 2011)
VibroTactor
(Hwang et al., 2012)
SoundWave
(Gupta et al., 2012)
Biomolecule detection
(Won et al., 2012)
WiSee
(Pu et al., 2013)
Medical Mirror
(Poh et al., 2011)
LIghtWave
(Gupata et al., 2011)
uTouch
(Chen et al., 2014)
m
m
m
e
t
w
c
e
m
e
t
c
microphone
EMI
touchscreen
camera
OTHER SOFTWARE APPROACH TO EMULATING INPUT MODALITIES
17. LIMITATION OF PREVIOUS WORKS
1. Limitation of Hardware augmentation approach
-They have to be attached to the device in the form of bulky accessories, which are seldom available to ordinary users (Maragos et al., 2008).
-The addition of hardware can mean additional production and maintenance costs for both device vendors and users.
2. Limitation of Software approach
-Some do not measure pressure applied by the user continuously(Esslet al., 2010; Heoet al., 2011b; Iwasaki et al., 2009)while others are fixed and limited with regard to the location of cameras (Kato et al., 2009; Sato et al.).
-They only focus on local problems. The general method for repurposing sensors have not yet investigated in HCI field.
19. PROPOSED CONCEPT : PSEUDOSENSOR
PseudoSensoris a sensor emulated
-using a (built-in) sensor or combination of (built-in) sensors
-for a different purpose from their original one (emulation of other functionality)
-without additional cost (sensors added)
PseudoSensor
Sensor
Sensor
20. PROPOSED CONCEPT : PSEUDOSENSOR
Our approach is to overcome the device constraints (e.g., absence or lack of sensors) by bypassing the effect the user create (e.g., pressure) to the emulated sensor (PseudoSensor).
Motor skill
Touchscreen
Human
Computer
Touch Effect
device
constraints
PseudoSensor
Pressure
Pressure Effect
Modalities, constraints, and effects (Obrenovic, Abascal, & Starcevic, 2007)
21. PROPOSED CONCEPT : PSEUDOSENSOR
Active vs. Passive
Event-based vs. Streaming-based vs. Recognition-based modality
Unimodalvs. Multimodal
Followed by the Simplified model of computing modalities
(Obrenovic, Abascal, & Starcevic, 2007)
PseudoSensor
Sensor
Sensor
24. The Right Number of Pressure Levels
-≤ 6 distinct levels (Ramos, Boulos, & Balakrishnan, 2004)
Selection Technique
-Dwell, Quick Release, Stroke, and Roll.
Feedback Design
-Continuous visual feedback is needed for a pressure widget (Ramos, Boulos, & Balakrishnan, 2004)
Perceptual Characteristics of Human Kinesthetic System
-The differential threshold of force is 7% -10% (over a force range of 0.5-200 N), while that of stiffness is 17% (Jones, 2000).
CONSIDERATIONS FOR DESIGNING PRESSURE INPUT TECHNIQUES
25. INSTANCES OF PSEUDOSENSOR
Eight applications to demonstrate how our approach supports pressure interaction for mobile devices.
a camera
a microphone
an accelerometer
a magnetometer
NailSenseand CamPress
MicPenand PseudoButton
ForceTouchand VibPress
MagGetzand MagPen
Pressure
Input
26.
27. PRESSURE ESTIMATION BY REPURPOSING A CAMERA
Two applications that repurpose a camera to emulate a pressure sensor
NailSenseisanovelinteractiontechniquethatrepurposesacamerasensortoemulateapressuresensor.Thistechniqueallowsuserstocontrolamobiledevicebyhoveringandslightlybending/extendingfingersbehindthedevice.Itdeterminesthepressureappliedwithauser’sfingertipbytrackingchangesincolorationofauser’sfingernailwithabuilt-incamera.
CamPressisanotherexamplethatdetectspressureassertedonamobilephonebyutilizinganinertialcameraonthemobiledevice.Ourtechniqueinferstheamountofpressureappliedontheinertialcameraofthemobiledevicebymeasuringtheluminanceofreflectedlight.
NailSense
CamPress
33. Magnetic traces for pressure-sensitive control widgets (MagGetz)
PRESSURE ESTIMATION BY REPURPOSING A MAGNETOMETER
wherep1isalastreferencepoint(e.g.,apointwhenthebuttonisfullypressed),p0isafirstreferencepoint(e.g., apointwhenthebuttonisnotpressed),andx’isavectorthatnewpointxprojectedontop0p1.
wherexisanewmagneticpointaccordingtotheuser’sinput,paisareferencepointofbuttonawithmaximumpressure,pbisamaximumpointofbuttonb,andp0isareferencepointofbuttonaandbwithminimumpressure.
Button a
Button b
Button a
38. RESULT : COMPLETION TIME (SEC)
Thecompletiontimeincreasesasinputlevelincreases.are,respectively,1.4(SD=0.34),1.73(SD=0.14), and2.30(SD=0.13)seconds.Therewerealsodifferencesincompletiontimeintermsofdifferentinputtechniques.TheaveragecompletiontimefordifferentinteractionsinL2was:0.91secondsforMicPen(SD=0.13);1.24secondsforVibPress(SD=0.04);1.31secondsforMagGetzButton(SD=0.07);1.38secondsforMagPen(SD=0.17);1.45secondsforPseudoButton(SD=0.32);1.5secondsforCamPress(SD=0.26);and2.03secondsforNailSense(SD=0.7)inanascendingsequence.Atwo-wayANOVArevealedsignificantdifferencesforcompletiontimeacrossinputlevels(F(2,153)=112.4,p<0.01), interactiontechniques(F(6,153)=29.5,p<0.01),andaninteractionwasfound(F(8,153)=5.98),p<0.01).
40. FINDINGS FROM THE EVALUTATION
The result is compellingwhen compared to previous works.
-We showed evidence that, with continuous visual feedback, users can reliably and quickly input using two to six pressure levels with accuracy ranging from 3.85% to 19.16% and from 1.4 to 2.3 secondsinteraction time, depending on the input levels (L2~L6).
-All techniques showed selection times similar to those obtained with specialized hardware (Cechanowiczet al., 2007; Ramos et al., 2004; Wilson et al., 2010) and MicPen, ForceTouch, VibPress, MagGetzButton, and MagPenshowed fewer errors than software pressure techniques (Goelet al., 2012; Heoet al., 2011b)
The errors and completion time varied depending on the number of input levels.
-Any subsequent increment took at least 1.2 times longer with considerably more errors.
The errors and completion time varied depending on the type of interactions.
-MagGetzButton (1.1%) vs. ForceTouch(20.3%)
-Repurposing a camera is cumbersome since it is easily affected by ambient light.
42. GRAMMAR OF PSEUDOSENSOR
HOW TO FORMALIZE THIS PSEUDOSENSOR?
Human.Motor<Pressure> → Computer.Microphone<Volume>
Attribute
Operator
Entity
Modality
Entity
Modality
43. Event ::= Entity[‘.’Modality]‘<’AttributeList’>’ (Operator Entity[‘.’Modality]‘<’AttributeList’>’)*
Entity ::= ‘Human’ | ‘Computer’ | ‘Environment’
Modality ::= ‘Microphone’ | ‘Accelerometer’ | ‘Camera’ | ‘Vibrator’ | ‘Motor’ | …
AttributeList::= Attribute (‘,’ Attribute)*
Attribute ::= ‘Gesture’ | ‘Pressure’ | ‘Color’ | ‘Light’ | ‘Movement’ | ‘Intensity’ | ‘Vibration’ | …
Operator ::= ‘→’ | ‘+data’ | ‘+feature’ | ‘+decision’
FORMALIZATION OF PSEUDOSENSOR
Extended Backus NaurForm (EBNF) grammar to define the syntax of PseudoSensor.
Human.motor<Touch, Pressure> → ( Computer.touchscreen<Touch> + Computer.accelerometer<Movement> )
Human touch and pressure affects touch attribute of a touchscreen and movement attribute of an accelerometer (ForceTouch)
44. Human.motor<Pressure> → Computer.accelerometer<Physical movement>
Human pressure affects the movement of accelerometer (repurposing an accelerometer).
Human.motor<Pressure> → ( Computer.speaker<Sound> → Computer.microphone<Spectrum, Volume> )
Human pressure affects the relationship in the way an inaudible sound from a speaker affects the spectrum and volume attributes of microphone (repurposing a microphone).
Human.motor<Touch, Pressure> → ( Computer.touchscreen<Contact position> +feature ( Magnet<Intensity> → Computer.magnetometer<Intensity> ) )
The human touch and pressure affects the contact position of touchscreen and the relationship in the way a purse of magnet affects the intensity of magnetometer
(repurposing a touchscreen and a magnetometer).
FORMALIZATION OF PSEUDOSENSOR
45. An overview of all applications we developed
AN OVERVIEW OF ALL APPLICATIONS WE DEVELOPED
46. An Overview of Software Approach to Enable Pressure Interaction
AN OVERVIEW OF PREVIOUS WORKS THAT ENBALE PRESSURE INPUT
48. SUMMARY OF DESIGN GUIDELINE FOR REPURPOSING SENSORS
1. PseudoSensorcan be empoweredby adding additional sensors
MicPencan be improved by compensating for sound energy variance according to the touch position and the speed of dragging measured by additional inertial sensor, a touchscreen
2. PseudoSensor, in certain aspects, sometimes demonstrate a better performance than a hardware-augmentation approach.
VibPresstechnique expands its interaction area beyond the touchscreen and enables in-pocket interaction, and the NailSensetechnique enables pressure-sensitive interaction in the air.
3. Active PseudoSensorusing public output channels may influence the performance of other systems that use the same technique.
PseudoButton, MagGetz, and MagPenuse public output channels (e.g., sound from a speaker and magnetism from a magnet) to build systems. Thus, we should consider the interference between systems for active Pseudosensorusing public channels.
50. CONCLUSION : SUMMARY OF STUDY
1.We presented a novel methodthat overcomes the limitations of hardware by emulating unavailable sensors through available input resources. This concept is unique in that it has not yet been introducedand summarized in the HCI field.
2.We presented a concrete set of applicationsand offered empirical evidence through a set of evaluations. The results of the evaluation are an important aspect of our work’s contribution to the field.
3.Through an exploration of both our own examples and the corpus of related work, we established a set of design guidelines. These unique guidelines can provide designers with flexible and reusable solutions when faced with insufficient sensor resources or suboptimal conditions.
51. 1.PseudoSensoris limited in terms of hardware settings.
Since PseudoSensorrelies on inertial sensors of the device, a limitation exists in terms of hardware settings: VibPresstechnique (compatibility & battery consumption issue), and PseudoButtontechnique (spatial constraints).
2. PseudoSensoraffects the performance of other systems.
When PseudoSensoris active and using public output channels, it may influence the performance of other systems: PseudoButton, MagGetz, and MagPen(interference issue).
3. PseudoSensormay cause negative effects on user experience.
Depending on a form of PseudoSensor, it may cause negative effect on user experience: VibPresstechnique (fatigue and noisy sound), and CamPresstechnique (privacy concerns).
4. There is no comparison with the previous hardware-augmentations.
We have empirically proved our method through various applications, but there is no comparison with the previous hardware-augmentation approaches. The comparison is needed.
CONCLUSION : LIMITATIONS
52. Expanding our vision by exploring a wider range of applications using our technique.
-Various modalities repurposed and sensor emulated (e.g., EMG, EEG, humidity)
-Different devices and domains (e.g., wearable, robot, and Internet of Things).
Conducting a longitudinal study and comparing them with corresponding real sensors.
More detailed investigation on the effect of repurposing sensors.
CONCLUSION : FUTURE WORK
Images from Google, Fin ring, Pebble, Samsung, and Amazon website (Nov. 26. 2014)