Paper Reading #24: Gesture Avatar

Gesture Avatar

by: Hao Lü and Yang Li


Authors
Researchers!
Lü does research at the University of Washington while Li (still) does research for Google Research.

Presentation Venue
"CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2011. (Yippee..)

Summary
Hypothesis
Lü hypothesized that their application, Gesture Avatar, would perform slower than competitor Shift on large targets, but faster on small targets.  Gesture Avatar would have fewer errors than Shift.  Walking vs. Sitting would not have much affect on Gesture Avatar.  It is basically a "look at these reasons why we're better than that guy" paper.

Methods
The application was developed for the Android and uses some relatively simple bounding box mathematics of proximity to determine what is a likely target for the intended avatar being created.  Users could then either use the target as desired, modify the avatar to the correct target, or dismiss it altogether.  To study the application in comparison with Shift, a study was conducted where half of the participants learned to use Shift, then Gesture Avatar, and the other half learned in the reverse order.  Their task was to select and manipulate targets in a small app based on different levels of ambiguity both while sitting and while walking.  Comparisons in variance were tested using ANOVA with p < 0.001.

Results
Each of Lü's three hypotheses were supported by the test's data.  Gesture Avatar was able to successfully manipulate targets after being selected, and did perform better than Shift on smaller targets with less errors, whether sitting or walking.

Discussion
I am glad to see that they were able to successfully approach a problem with a solution.  Making small targets larger seems like such an obvious thing, but apparently a study needed to be conducted to decide what to do about it.  There are still obvious issues with this approach, though.  The main thing is the need to switch into a gesture mode to perform the desired task.  The other thing that bothers me is its applicability to a situation.  To turn on the application, draw an avatar, and then ensure that it's connected to the correct target, I could probably have fumbled with my fat fingers and managed the control already before I found it with the gesture.  This is particularly true for something like a hyperlink on a website.  In a browser, all I have to do is make a quick flick of my fingers to zoom in, and the hyperlink is suddenly very easy to touch.  Additionally, Gesture Avatars maintained that the avatars could be completely arbitrary... meaning if I have some gesture I want to make, I have to tell it which target to bind to.  Not only this, but the avatar is not remembered.  I have to draw it again if I need it again after clicking a hyperlink and moving to a new page with the same widgets to control.  With some more study, this could lead in the right direction for eliminating ambiguity of fat fingers, though.

Paper Reading #22: Mid-air Pan-and-Zoom on Wall-sized Displays

Mid-air Pan-and-Zoom on Wall-sized Displays

by: Mathieu Nancel, Julie Wagner, Emmanuel Pietriga, Olivier Chapuis, and Wendy Mackay


Authors
Researchers!
All research was performed at the University of Paris for INRIA (taglined as Inventors for the digital world).

Presentation Venue
"CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2011. (Again...The UIST has been repalced..)

Summary
Hypothesis
Nancel proposed several hypotheses which were really just interpretations from many previous studies, such as user preference for linear gestures over circular gestures, and two handed over one handed gestures.  Nancel also hypothesized that too many degrees of freedom would prove too tiresome.

Methods
Participants were presented with a large system which they were required to stand different distances from and attempt to perform the gestures asked of them.  The sessions were split up to prevent fatigue and boredom because the process was long and required some potentially unnatural interactions.  The different actions were compared using standard ANOVA tests.

Results
Nancel found support for his hypotheses quantitatively as well as qualitatively.  Participants suggested that two handed gestures were easier because they did not require the random jumbled hand gestures that come about because of single hand gestures.  Linear panning and zooming was preferred by the participants as well.

Discussion
I read this paper with an eyebrow raised and a constant question in the back of my mind: "So, what?"  They took a whole bunch of previous research and combined it.  Sure, that's what almost every research endeavor is, but this didn't even feel like it had an innovative twist.  It was more like it was simply confirming what everyone else said in a single place with one final confirmation experiment.  On top of that, how often do we really use wall sized displays?  This might be more meaningful in the future when I'm walking down the street and the entire wall of the building next to me is a streamline of advertisements that zoom in when I get closer and actually try to take a look at them.  But, that means it's more important for marketing and sales.  To me, as a consumer, it's a nuisance.  Kudos for successfully proving your hypotheses though, Nancel.

Paper Reading #23: User-defined motion gestures

User-defined Motion Gestures for Mobile Interactions

by: Jaime Ruiz, Yang Li, and Edward Lank


Authors
Researchers!
Ruiz and Lank do research at the University of Waterloo, while Li researches with Google Research.

Presentation Venue
"CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2011. (Again...The UIST has been repalced..)

Summary
Hypothesis
Ruiz hypothesized that there are a set of common motion gestures that users would find the most natural and consistent mappings to interactions with a mobile device.

Methods
Ruiz put custom software on an Android so that it would record accelerometer data and lock the screen to prevent any feedback to the user.  Twenty participants were chosen and asked to create several gestures for a set of tasks that needed to be performed on the phone, such as answering a call.  Data about the gesture itself was actually sequentially recorded on a computer connected to the smartphone.  Participants were then interviewed to analyze some of their decisions in creating gestures for the phone.

Results
Ruiz found that there were indeed several common gestures that people found.  The most common gesture was lifting the phone to the ear in order to answer the phone because it felt the most natural.  It was also very common for them to mimic replacing the phone on an old receiver in order to end a call.  Most participants expressed a desire to have the same gesture available for multiple analogous behaviors (next, previous in different applications).

Discussion
I find myself asking why this sort of research was not performed sooner.  Obviously, something like answering a phone by bringing it to your ear is how many of us grew up answering a phone.  It is far more natural than pressing a button (Flip phones are better in this regard, in my opinion).  With this in mind, I am not particularly impressed with the research discovery because it is semi obvious. But, I am relieved that SOMEONE did it, so maybe the data can be incorporated.  However, there are still developing methods of handling data on smartphones.  Just because we CAN perform motion gestures, that does not necessarily mean that we need to.  I personally feel that moving my phone to the left or right in order to proceed to the next or previous is far less natural that swiping the application with my own finger.  This is particularly true with images.  If I am interacting with many real photos, I am not likely to pick up the table and shake it to look at the next image.  I am much more likely to touch the picture, move it, and bring the next one into focus.

Paper Reading #16: Classroom-based Assistive Technology

Classroom-based Assistive Technology: Collective Use of Interactive Visual Schedules by Students with Autism

by: Meg Cramer,Sen H. Hirano, Monica Tentori, Michael T. Yeganyan and Gillian R. Hayes


Authors
Researchers!
Cramer, Hirano, Tentori, Yeganyan, and Hayes all do research at the University of California.

Presentation Venue
"CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2011. (Hurray for leaving the UIST!)

Summary


vSked is an assistive technology for educating children with autism.  The paper focuses on a case study of the technology used over a year in autistic classrooms.

Hypothesis
Cramer hypothesizes that vSked can be used to create new community learning opportunities and facilitate streamlined teacher systems for aiding the children.

Methods
vSked has a large monitor that is placed at the front of a classroom which keeps track of the schedule of activities for the day.  It also displays the progress of each child within the current activities.  Children each had a touch device with their information, an image reminding them of their incentive for completing tasks, and options to choose from in their activities.  The technology was used for one year in a classroom and observed for over 200 hours plus a few interviews with the teachers and children.

Results
Teachers commented that the system made their jobs much easier.  They felt like they could spend more time with each of the students.  Some students appeared to be more motivated to accomplish tasks.  Students did not require as much prompting from teachers to complete their tasks.  When students earned the fireworks display for earning their incentives, a new community opportunity was observed as students cheered at their success.

Discussion
I have extremely mixed feelings about this research.  On the one hand, it is very cool that we are getting more technology to help teach autistic children.  I have serious reservations about the use of the technology, though.  Firstly, this paper says it focuses on the classroom as a whole community and takes emphasis off of the individual children.  The students are given computers which they must interact with instead of one another.  I have a friend with an autistic child, and I talked to her about her insights on technology assistance.  Before hearing her opinions though, I suggested that the technology might take the student's focus away from the community and their social interactions.  Specifically, I wanted to know if the student being more motivated was ACTUALLY more motivated, or if he was just more comfortable working with a computer than he was comfortable interacting with people.  My friend echoed my thoughts by informing that her son would be perfectly happy if he could just play on the computer all day instead of play with other kids.  Additionally, she pointed out that perceived progress in the class due to the technology's streamlining effort for teachers would actually just result in the school administration giving them more to do because on paper they appear to be progressing - regardless of actual progress made.  The technology is awesome, and I like the idea of helping, but I fear that relying on technology to help TEACH them ignores the still persistent problem (and perhaps impairs) of no social skills.  Specifically, she mentioned things we take for granted have to be actively considered by an autistic child.  Thus, when we detect sarcasm, an autistic child must think about it or it is unperceived.  Text on a computer cannot convey something such as sarcasm at all.

Paper Reading #17: Privacy Risks in Wearable Sensors

Privacy risks emerging from the adoption of innocuous wearable sensors in the mobile environment

by: Andrew Raij, Animikh Ghosh, Santosh Kumar, Mani Srivastava


Authors
Researchers!
Raij - University of South Florida
Ghosh - SETLabs
Kumar - University of Memphis
Srivastava - University of California

Presentation Venue
"CHI '11 Proceedings of the 2011 annual conference on Human factors in computing systems" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2011. (Finally! A new venue!)

Summary


Study was conducted using sensor data in mobile devices to determine what level of risk users perceived about the information which could be gleaned from these sensors about themselves.

Hypothesis
Raij hypothesized that people would have a higher level of concern for information that could be released from a worn sensory study when they had observed a personal stake in the information to be released.  Raij also wished to see what manipulations of data would increase/decrease concern for information.

Methods
A user study was conducted with two groups, a control group that took an information privacy survey, and the experiment group which wore sensory capturing devices for three consecutive days, took the privacy survey, saw the results from their sensory data, and then took the privacy survey again.

Results
Participants who took the survey had similar levels of concern as did participants who wore the sensor devices in their first survey.  Concern rose drastically in the second survey when the participants saw what information about themselves could be derived from sensory information.  Physical location did not display an increase in concern.  It was suggested that this was because participants expected their location to be trackable by the very nature of the devices.  Emotional states (stress) and conversation had the highest levels of concern, especially when coincided with time stamps because that allowed one to determine things such as who one was talking to and when they had a seizure.

Discussion
This study demonstrated to me that people probably don't think deeply enough about situations which they enter into.  The participants with the sensory data were presented with disclosures before wearing the devices so they knew what could be obtained from the devices.  I acknowledge that they did not understand the extent to which this information could be accrued, but things such as time stamps should have been recognizable as problems.  The very good thing about this survey is that it allows us to now know for the future that such subtle cues can be used to gather information, and we as users should be more conscious of what information without abstractions we allow.

Paper Reading #15: Madgets

Madgets: Actuating Widgets on Interactive Tabletops

by: Malte Weiss, Florian Schwarz, Simon Jakubowski and Jan Borchers


Authors
Seeing as how I don't read German... I find it adequate to simply say that these authors performed their research in association with the RWTH Aachen University in Germany.

Presentation Venue
"UIST '10 Proceedings of the 23nd annual ACM symposium on User interface software and technology" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2010.

Summary


Madgets explores a new approach to moving widgets on an interactive tabletop.  It creates widgets with magnets and manipulates several electromagnets to facilitate interaction with those widgets from the tabletop.

Hypothesis
Weiss hypothesis presents Madgets as a new approach to widgets on interactive tabletops.  Weiss suppose that variable electromagnets in conjunction with deterministic magnets on widgets can be used to simulate interactions with their algorithms.

Methods
Using a whole bunch of mathematics and fancy technology, Weiss constructed an interactive tabletop that has many electromagnets on the bottom and can interact with arbitrary magnetic widgets placed on it.  Users are able to perform some actions with the widgets, manipulate them, and have some 3D interactions with them as well.  Actually moving the widgets goes through yet more math and some physics.  There appears to lack user studies thus far.

Results
Obviously, the tabletop system was created.  Widgets can be moved around and interacted with on the tabletop.  Weiss reports that there is a great deal of time involved in setup and registration with the system, such as approximately two hours for making the system aware of the new widget, and an hour to actually create the new physical widget itself.

Discussion
While this design space is interesting and might have potential in some future work scope, I presently do not see it and do not find it too interesting.  It's presently at too basic of a stage for me to be excited about what implications it might have on tomorrow's gadgets.  What use do I have for a 3D-able "check-box" that I have to spend 5+ hours making, and then it's the size of my hand anyways?  When and if these get scaled down and start having more realistic applications, I may be more interested - particularly in the mechanical sound feedback.  Synthesized audio is nice, but it always lacks that slight imperfection and personality of a natural sound.

Paper Reading #14: TeslaTouch

TeslaTouch: Electrovibration for Touch Surfaces

by: Olivier Bau, Ivan Poupyrev, Ali Israr, and Chris Harrison


Authors
All authors research at Disney Research Pittsburg, in Pittsburg, PA.  Bau received his PhD from a university in Paris.

Presentation Venue
"UIST '10 Proceedings of the 23nd annual ACM symposium on User interface software and technology" as per the document specs from the ACM digital library, dl.acm.org. The presentation took place in New York City, NY, 2010.

Summary


TeslaTouch introduces a technology that uses a phenomenon discovered in the 50's called electrovibration to provide tactile feedback for users on interfaces.  This paper introduces TeslaTouch, explains its studies, and analyzes pros and cons of electrovibration vs. mechanical vibrotactile feedback.

Hypothesis
As a new technology, Bau's main hypothesis is that electrovibration is a viable option for tactile feedback on interfaces and implements it as TeslaTouch.

Methods
TeslaTouch is set up with an understanding of a bunch of technical jargon that basically means small electric shocks are used to create a sensation similar to different surfaces that a finger might touch as it moves across the surface.  Users were tested using the surface to determine thresholds for voltage perception, and other things such as apparent smoothness/stickiness of the surface.

Results
TeslaTouch was successfully created and tests suggested that users were successfully able to distinguish between their criteria of sensations.  It was also discovered the users could feel quite low voltage which could serve for mobile device power conservation.  It was observed that electrovibration tactile feedback had many advantages over mechanical vibrotactile feedback - chiefly coming from the fact that electrovibbration feedback requires no moving components.

Discussion
I never even considered this type of interface coming out... but now that I've read about it, I'm intrigued, vaguely think "duh" and would like to test it out myself.  While I know that such feedback in my phone is not necessary by any means, I do believe it would have the "wow" factor that is missing from many new technologies and applications for my phone.  It would be pretty awesome to be able to send a text that says, "here, it felt something like this."  The limitation to a single moving finger is a bit of a let down, however.  I understand that it is based on friction-imitation, so movement is necessary, but restricting it to a single finger just seems harsh.  This certainly makes me wonder if we're going to start having more fake fruit that feels like real fruit in every way except perhaps weight (and maybe even in weight...).