We present empirical results about users’ gesture preferences
for smart rings by analyzing 672 gestures from 24 participants.
We report an overall low consensus (mean :112, maximum
:225 on the unit scale) between participants’ gesture
proposals, and we point to the challenges of designing highly generalizable ring gestures across users. We also contribute
to the practice of gesture elicitation studies by discussing how
a priori conditions (e.g., participants’ traits, such as creativity
and motor skills), commitment and behavior during the
experiment (e.g., their thinking times), but also a posteriori
aspects (the experimenter’s choice of criteria to group gestures
into categories) affect agreement. We offer design guidelines
for ring gestures informed by our empirical observations, and
present a collection of gestures reflective of our participants’
mental models for effecting commands using smart rings.
The Codex of Business Writing Software for Real-World Solutions 2.pptx
Gestures for Smart Rings: Empirical Results, Insights, and Design Implications
1. Gestures for Smart Rings:
Empirical Results, Insights, and
Design Implications
Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
2. Gestures for Smart Rings:
Empirical Results, Insights, and
Design Implications
Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Growing interest in wearables and
miniaturization, including smart rings
3. Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Gestures for Smart Rings:
Empirical Results, Insights, and
Design ImplicationsGrowing interest in wearables and
miniaturization, including smart rings
Most efforts on technology and classification
algorithms to recognize ring gestures effectively
4. Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Gestures for Smart Rings:
Empirical Results, Insights, and
Design ImplicationsGrowing interest in wearables and
miniaturization, including smart rings
Most efforts on technology and classification
algorithms to recognize ring gestures effectively
However, users’ preferences for smart ring
gestures have not been examined so far
5. Gestures for Smart Rings:
Empirical Results, Insights, and
Design Implications
Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Our contributions:
6. Gestures for Smart Rings:
Empirical Results, Insights, and
Design Implications
Our contributions:
Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
The first scientific investigation to collect, analyze, and
understand users’ preferences for ring gestures
7. Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Gestures for Smart Rings:
Empirical Results, Insights, and
Design Implications
Bogdan-Florin Gheran
University of Suceava,
Romania
The first scientific investigation to collect, analyze, and
understand users’ preferences for ring gestures
Practical results:
1. Analysis of agreement
2. Generic taxonomy for ring gestures with 5 dimensions
(nature, structure, complexity, symmetry, and locale) to
analyze and inform ring gesture designs
3. Collection of representative gestures
4. Practical recommendations for designing ring gesture
commands
Our contributions:
9. Elicitation study
1. Present participants
with the effect of a
command
2. Ask participants to
propose a gesture
for the command
3. Perform agreement
analysis
1
2
3
Wobbrock et al., CHI 2005, CHI 2009
Vatavu and Wobbrock, CHI 2015
10. Elicitation study
1. Present participants
with the effect of a
command
2. Ask participants to
propose a gesture
for the command
3. Perform agreement
analysis
1
2
3
24 participants
11. Elicitation study
24 participants
No. Referent
1 Turn the TV on/off
2 Start player
3 Turn the volume up
Turn the volume down4
5 Next
6 Previous
7 Turn AC on/off
8 Turn lights on/off
9
10
11
12
13
14
Brighten lights
Dim lights
Turn heat on/off
Turn alarm on=off
Answer phone call
End phone callx 14 referents
12. Elicitation study
24 participants
= 672 gesture proposals
x 2 rings (Ring Zero)
x 14 referents
No. Referent
1 Turn the TV on/off
2 Start player
3 Turn the volume up
Turn the volume down4
5 Next
6 Previous
7 Turn AC on/off
8 Turn lights on/off
9
10
11
12
13
14
Brighten lights
Dim lights
Turn heat on/off
Turn alarm on=off
Answer phone call
End phone call
13. Elicitation study
24 participants
x 2 rings
x 14 referents
No. Referent
1 Turn the TV on/off
2 Start player
3 Turn the volume up
Turn the volume down4
5 Next
6 Previous
7 Turn AC on/off
8 Turn lights on/off
9
10
11
12
13
14
Brighten lights
Dim lights
Turn heat on/off
Turn alarm on=off
Answer phone call
End phone call
Agreement rate per
referent ∈ [0. . 1]
14. No. Referent
1 Turn the TV on/off
2 Start player
3 Turn the volume up
Turn the volume down4
5 Next
6 Previous
7 Turn AC on/off
8 Turn lights on/off
9
10
11
12
13
14
Brighten lights
Dim lights
Turn heat on/off
Turn alarm on=off
Answer phone call
End phone call
Elicitation study
24 participants
Agreement rate per
referent ∈ [0. . 1]
Agreement rate
formula (Vatavu and
Wobbrock, CHI 2015):
15. Elicitation study
2
3
24 participants
x 14 referents
= 672 gesture proposals
x 2 rings
Agreement rate per
referent ∈ [0. . 1]
The formula of Vatavu and
Wobbrock:
22. Motion seems to be preferred over static
gestures
Users prefer “simple” gestures performed with
the dominant hand
1
2
23. Motion seems to be preferred over static
gestures
Users prefer “simple” gestures performed with
the dominant hand
Consider gesture designs that exploit the
surface of the ring and/or its unique shape and
form factor
1
2
3
24. Motion seems to be preferred over static
gestures
Users prefer “simple” gestures performed with
the dominant hand
Consider gesture designs that exploit the
surface of the ring and/or its unique shape and
form factor
Reuse touch gestures and touch-based
interaction paradigms from smartphones due
to their familiarity to users
1
2
3
4
25. Consider multiple gestures for the same
command (many-to-one association), but also the
same gesture for many commands (one-to-many),
if permitted by context
5
26. Consider multiple gestures for the same
command (many-to-one association), but also the
same gesture for many commands (one-to-many),
if permitted by context
Favor symmetry in designing bimanual gestures
5
6
27. Consider multiple gestures for the same
command (many-to-one association), but also the
same gesture for many commands (one-to-many),
if permitted by context
Favor symmetry in designing bimanual gestures
Consider gestures for operating imaginary objects
in mid-air
5
6
7
28. Consider multiple gestures for the same
command (many-to-one association), but also the
same gesture for many commands (one-to-many),
if permitted by context
Favor symmetry in designing bimanual gestures
Consider gestures for operating imaginary objects
in mid-air
Design that fosters synergy with body gestures
5
6
7
8
29. Consider multiple gestures for the same
command (many-to-one association), but also the
same gesture for many commands (one-to-many),
if permitted by context
Favor symmetry in designing bimanual gestures
Consider gestures for operating imaginary objects
in mid-air
Design that fosters synergy with body gestures
Design ring gestures that foster multimodal input
5
6
9
7
8
30. Conclusion
The first examination of users’ preferences for
smart ring gestures
Taxonomy to analyze ring gestures
Collection of recommended gesture types
Set of guidelines for designing smart ring gestures
31. Thank you for your attention!
Radu-Daniel Vatavu
University of Suceava, Romania
Bogdan-Florin Gheran
University of Suceava, Romania
Jean Vanderdonckt
Université catholique de Louvain, Belgium
Editor's Notes
Good morning, everyone! Thank you for coming to this session. This work is about gestures performed using smart rings, and especially about users’ preferences for gestures they would actually use to effect commands with one or two smart rings. I am a Phd Student at the University of Suceava and this work was done in collaboration with Jean Vanderdonckt from UCL and Radu-Daniel Vatavu from USV
There has been a growing interest in designing and prototyping wearables in our community, such as smart rings. But the largest efforts went into technology, especially miniaturization, sensing, communications, and so on.
Also, a lot of effort has been put into recognizing smart ring gestures accurately, and several papers have proposed gesture recognition approaches for this purpose.
However, an important aspect has been neglected so far. In this work, we look at users’ preferences for gestures performed using smart rings in order to understand mental models for this task and to develop recommendations and guidelines to inform gesture set design for smart ring gesture interfaces.
We make several contributions in this paper:
First, we provide the first scientific investigation of user preferences for smart ring gestures, for which we conducted a user study to elicit users’ preferences for gestures performed using one and two smart rings.
There are many practical results. First, we provide a detailed analysis of the consensus reached by our participants with respect to their preferences for ring gestures they would use to effect various tasks. Second, we conducted a classification of all elicited gestures according to the categories of a new taxonomy smart ring gestures with 5 dimensions (…). This taxonomy is useful to understand various performance aspects of ring gestures (as we did in the paper), but it may also be used to inform gesture set design. Third, we compile a set of representative gesture commands to effect various tasks in a smart home environment, such as turning on the TV set or turning off the lights. Fourth, we elaborate on design guidelines and recommendations to help practitioners in our community to design gestures for smart rings based on our insights and observations from our empirical study.
We conducted a gesture elicitation study to collect users’ preferences for smart ring gestures they would like to use.
The gesture elicitation methodology was introduced by Wobbrock and his collegues in 2005 and 2009 and the updated methodology was published in 2015 by Vatavu and Wobbrock. The methodology is very simple to apply in practice and this is probably why it has been largely adopted in our community. Since 2009, there have been over 100 studies and papers published on gesture elicitation for various gesture technology and contexts of use. Even yesterday, we saw a very beautiful work by Tang and his collegues on gestures for augmented reality. There are three stages in applying the gesture elicitation methodology. First, you gather a group of participants and present them with a set of tasks and the effect of those tasks such as turning on the TV or turning off the lights in a smart home. Then, you ask each participant to think of a suitable gesture, using the smart rings in our case, to effect each task. At then end, once all participants’ gestures were collected, you perform a so-called analysis of agreement to understand the level of consensus or agreement that is present in the data. The gestures that receive the highest level of agreement can be recommended for use in practice.
We had 24 participants in our study and we considered 14 referents for a smart home scenario, such as 1, 10, or 12.
We used two Ring ZERO devices. Let me give you a live demonstration of the kind of data they report. These rings embed … explicat tehnologia cat demonstrezi
In total, we collected 672 gestures from our participants.
To evaluate agreement numerically, we computed the agreement rate for each referent
Using the formula of Vatavu and Wobbrock. (In this formula, where n is the number of participants from which gesturesare elicited, and di; j evaluates to 1 if the i-th and j-th participants are in agreement over referent r and to 0 otherwise.) A nice thing about the agreement rate is that is falls in the [0,1] interval, where 0 means no agreement at all and 1 means perfect agreement (all participants proposed the same gesture).
These are a few examples of the gestures proposed by our participants using one or both rings. A link to the video is also available in our paper and the video is on youtube. All these gestures were grouped by similarity into clusters of similar types and analyzed according to a new taxonomy for ring gestures that we proposed in this work. As you can see, participants were free to propose any gestures they wished and were not constrained by the technical features or limitations of the rings. We wanted participants to believe that any gesture was possible.
There are three practical results of our gesture elicitation study:1, 2, and 3.
In terms of agreement rates, we observed that agreement was overall low. The highest agreement was observed for the Next referent (which means advance to the next item in a list). The value of the maximum agreement was about 0.200, which is nevertheless low compared to other elicitation studies published in the literature. The level of agreement of gestures proposed with two smart rings instead of one is even lower. In our paper, we discuss reasons for observing such low agreement levels, such as our choice of criteria to group gestures into clusters of similar types or the novelty of the smart ring technology. Please see the paper for detailed discussion.
Despite the low agreement, we hand picked a few gestures proposed by our participants that we believed are interesting to reveal mental models of interaction.
In this figure, you can see a selection of such gestures. At the top, we exemplify gestures performed with one smart ring, which participants worn on their dominant hand. At the bottom, you can see gestures performed with two smart rings. You can see a large variety of gesture types, such as hand motion, finger gestures in mid-air, hand poses, touches and taps on the ring surface, pointing in mid-air or even some users that imagined holding virtual objects, such as a TV remove control. For gestures performed with two smart rings, the variety was even larger. You can see hands rubbing, rotating the ring around the finger, clapping, tapping both rings or performing sequences of taps, and even some unusual actions, such as blowing on the ring. Please see our paper for detailed discussion of such interesting gestures invented by our participants.
Another important outcome of our study is a set of design guidelines and recommendations that we offer to practitioners to inform gesture design for smart ring user interfaces.
First, we observed that the large majority of gestures proposed by our participants involved motion, so motion seems to be …
We also observed a preference for simple gestures involving little movement or effort and that were mostly performed with the dominant hand
Some participants made use of the surface of the ring to propose gestures and the ring has a unique form factor that can be exploited to accommodate a variety of interesting interactions, such as tapping the ring at various locations, rotating the ring around the finger, pulling the ring out and putting it back on the finger or combinations of these.
We also observed several gesture types that show the familiarity of our participants with touch gesture user interfaces. Therefore, one recommendation is to exploit this user knowledge to accommodate a easy transition from touch interaction to smart ring interaction.
We can also recommend associating multiple gestures to execute the same command. The viceversa is also an option, but only if your context information can be used to clearly identify the task that the user intends to execute.
Most of the gestures that we observed were symmetrical in nature so we can recommend to favor …
Some participants imagined objects that they operated using hand poses, such as one participant turned on and off the TV set using an imaginary TV remove on which they pressed a button. We can recommend future explorations in this direction.
Not all gestures proposed by participants could be recognized by smart rings. Some of them in fact require other sensors, such as whole-body gesture sensors. Even if today smart ring technology cannot sense some of the gestures that we describe in this work, there is an opportunity to design hybrid gesture interfaces that combine smart rings with other sensors, such as video cameras to design rich interactions.
But also consider multimodal input. To give you an example, one participant imagined that the smart ring embedded a microphone, so they spoke into the ring, while other participant blew on the ring. So, there is also potential for interesting multimodal user interfaces to be explored involving not just gestures, but other input modalities as well.
In conclusion, we performed the first examination…. In this process, we proposed a new taxonomy to analyze ring gestures (that I did not cover in this presentation, but please look at the paper where many interesting results are presented). We compiled a collection of interesting gesture types for user interfaces for one and two smart rings and we used our empirical observations to derived a set of guidelines to inform design of gestures for smart rings. We hope that our results will be useful for the community to design gestures for smart rings informed by actual user preferences and reflective of their actual behavior.
Thank you very much for your attention and I am happy now to take your questions.