Category Archives: multi-modal input

ACM SIGMOBILE 2017 Test of Time Award for “Sensing Techniques for Mobile Interaction”

Recently the SIGMOBILE community recognized my turn-of-the-century research on mobile sensing techniques with one of their 2017 Test of Time Awards.

This was the paper (“Sensing Techniques for Mobile Interaction“) that first introduced techniques such as automatic screen rotation and raise-to-wake to mobile computing — features now taken for granted on the iPhones and tablets of the world.

The award committee motivated the award as follows:

This paper showed how combinations of simple sensors could be used to create rich mobile interactions that are now commonplace in mobile devices today. It also opened up people’s imaginations about how we could interact with mobile devices in the future, inspiring a wide range of research on sensor-based interaction techniques.

SIGMOBILE-Test-of-Time-Award-photo-700h

And so as not to miss the opportunity to have fun with the occasion, in the following video I reflected (at times irreverently) on the work — including  what I really thought about it at the time I was doing the research.

And some of the things that still surprise me about it after all these years.

You can find the original paper here.


Thumbnail - Ken Hinckley CHI Academy 2014 InducteeHinckley, K., Pierce, J., Sinclair, M., Horvitz, E. ACM SIGMOBILE 2017 Test of Time Award. [SIGMOBILE Test of Time Awards archive]

Paper: WritLarge: Ink Unleashed by Unified Scope, Action, & Zoom

Electronic whiteboards remain surprisingly difficult to use in the context of creativity support and design.

A key problem is that once a designer places strokes and reference images on a canvas, actually doing anything useful with key parts of that content involves numerous steps.

Hence, with digital ink, scope—that is, selection of content—is a central concern, yet current approaches often require encircling ink with a lengthy lasso, if not switching modes via round-trips to the far-off edges of the display.

Only then can the user take action, such as to copy, refine, or re-interpret their informal work-in-progress.

Such is the stilted nature of selection and action in the digital world.

But it need not be so.

By contrast, consider an everyday manual task such as sandpapering a piece of woodwork to hew off its rough edges. Here, we use our hands to grasp and bring to the fore—that is, select—the portion of the work-object—the wood—that we want to refine.

And because we are working with a tool—the sandpaper—the hand employed for this ‘selection’ sub-task is typically the non-preferred one, which skillfully manipulates the frame-of-reference for the subsequent ‘action’ of sanding, a complementary sub-task articulated by the preferred hand.

Therefore, in contrast to the disjoint subtasks foisted on us by most interactions with computers, the above example shows how complementary manual activities lend a sense of flow that “chunks” selection and action into a continuous selection-action phrase. By manipulating the workspace, the off-hand shifts the context of the actions to be applied, while the preferred hand brings different tools to bear—such as sandpaper, file, or chisel—as necessary.

The main goal of the WritLarge project, then, is to demonstrate similar continuity of action for electronic whiteboards. This motivated free-flowing, close-at-hand techniques to afford unification of selection and action via bimanual pen+touch interaction.

WriteLarge-hero-figure

Accordingly, we designed WritLarge so that user can simply gesture as follows:

With the thumb and forefinger of the non-preferred hand, just frame a portion of the canvas.

And, unlike many other approaches to “handwriting recognition,” this approach to selecting key portions of an electronic whiteboard leaves the user in complete control of what gets recognized—as well as when recognition occurs—so as not to break the flow of creative work.

Indeed, building on this foundation, we designed ways to shift between flexible representations of freeform content by simply moving the pen along semantic, structural, and temporal axes of movement.

See our demo reel below for some jaw-dropping demonstrations of the possibilities for digital ink opened up by this approach.

Watch WritLarge: Ink Unleashed by Unified Scope, Action, and Zoom video on YouTube


WritLarge-CHI-2017-thumbHaijun Xia, Ken Hinckley, Michel Pahud, Xioa Tu, and Bill Buxton. 2017. WritLarge: Ink Unleashed by Unified Scope, Action, and Zoom. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI’17). ACM, New York, NY, USA, pp. 3227-3240. Denver, Colorado, United States, May 6-11, 2017. Honorable Mention Award (top 5% of papers).
https://doi.org/10.1145/3025453.3025664

[PDF] [Watch 30 second preview on YouTube]

Paper: Thumb + Pen Interaction on Tablets

Modern tablets support simultaneous pen and touch input, but it remains unclear how to best leverage this capability for bimanual input when the nonpreferred hand holds the tablet.

We explore Thumb + Pen interactions that support simultaneous pen and touch interaction, with both hands, in such situations. Our approach engages the thumb of the device-holding hand, such that the thumb interacts with the touch screen in an indirect manner, thereby complementing the direct input provided by the preferred hand.

For instance, the thumb can determine how pen actions (articulated with the opposite hand) are interpreted.

thumb-pen-fullsize

Alternatively, the pen can point at an object, while the thumb manipulates one or more of its parameters through indirect touch.

Our techniques integrate concepts in a novel way that derive from radial menus (also known as marking menus) and spring-loaded modes maintained by muscular tension — as well as indirect input, and in ways that leverage multi-touch conventions.

Our overall approach takes the form of a set of probes, each representing a meaningfully distinct class of application. They serve as an initial exploration of the design space at a level which will help determine the feasibility of supporting bimanual interaction in such contexts, and the viability of the Thumb + Pen techniques in so doing.

Watch Thumb + Pen Interaction on Tablets video on YouTube


thumb-pen-thumbKen Pfeuffer, Ken Hinckley, Michel Pahud, and Bill Buxton. 2017. Thumb + Pen Interaction on Tablets. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI’17). ACM, New York, NY, USA, pp. 3254-3266 . Denver, Colorado, United States, May 6-11, 2017.
https://doi.org/10.1145/3025453.3025567

[PDF] [Watch 30 second preview on YouTube]

Paper: As We May Ink? Learning from Everyday Analog Pen Use to Improve Digital Ink Experiences

This work sheds light on gaps and discrepancies between the experiences afforded by analog pens and their digital counterparts.

Despite the long history (and recent renaissance) of digital pens, the literature still lacks a comprehensive survey of what types of marks people make and what motivates them to use ink—both analog and digital—in daily life.

As-We-May-Ink-fullsize

To capture the diversity of inking behaviors and tease out the unique affordances of pen-and ink, we conducted a diary study with 26 participants from diverse backgrounds.

From analysis of 493 diary entries we identified 8 analog pen-and-ink activities, and 9 affordances of pens. We contextualized and contrasted these findings using a survey with 1,633 respondents and a follow-up diary study with 30 participants, observing digital pens.

Our analysis revealed many gaps and research opportunities based on pen affordances not yet fully explored in the literature.


As-We-May-Ink-CHI-2017-thumbYann Riche, Nathalie Henry Rich, Ken Hinckley, Sarah Fuelling, Sarah Williams, and Sheri Panabaker. 2017. As We May Ink? Learning from Everyday Analog Pen Use to Improve Digital Ink Experiences. In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems (CHI’17). ACM, New York, NY, USA, pp. 3241-3253. Denver, Colorado, United States, May 6-11, 2017.
https://doi.org/10.1145/3025453.3025716

[PDF] [CHI 2017 Talk Slides (PowerPoint)]

Paper: Wearables as Context for Guiard-abiding Bimanual Touch

This particular paper has a rather academic-sounding title, but at its heart it makes a very simple and interesting observation regarding touch that any user of touch-screen technology can perhaps appreciate.

The irony is this: when interaction designers talk about “natural” interaction, they often have touch input in mind. And so people tend to take that for granted. What could be simpler than placing a finger — or with the modern miracle of multi-touch, multiple fingers — on a display?

And indeed, an entire industry of devices and form-factors — everything from phones, tablets, drafting-tables, all the way up to large wall displays — has arisen from this assumption.

Yet, if we unpack “touch” as it’s currently realized on most touchscreens, we can see that it remains very much a poor man’s version of natural human touch.

For example, on a large electronic-whiteboard such as the 84″ Surface Hub, multiple people can work upon the display at the same time. And it feels natural to employ both hands — as one often does in a wide assortment of everyday manual activities, such as indicating a point on a whiteboard with your off-hand as you emphasize the same point with the marker (or electronic pen).

Yet much of this richness — obvious to anyone observing a colleague at a whiteboard — represents context that is completely lost with “touch” as manifest in the vast majority of existing touch-screen devices.

For example:

  • Who is touching the display?
  • Are they touching the display with one hand, or two?
  • And if two hands, which of the multiple touch-events generated come from the right hand, and which come from the left?

Well, when dealing with input to computers, the all-too-common answer from the interaction designer is a shrug, a mumbled “who the heck knows,” and a litany of assumptions built into the user interface to try and paper over the resulting ambiguities, especially when the two factors (which user, and which hand) compound one another.

The result is that such issues tend to get swept under the rug, and hardly anybody ever mentions them.

But the first step towards a solution is recognizing that we have a problem.

This paper explores the implications of one particular solution that we have prototyped, namely leveraging wearable devices on the user’s body as sensors that can augment the richness of touch events.

A fitness band worn on the non-preferred hand, for example, can sense the impulse resulting from making finger-contact with a display through its embedded motion sensors (accelerometers and gyros). If the fitness band and the display exchange information and id’s, the touch-event generated can then be associated with the left hand of a particular user. The inputs of multiple users instrumented in this manner can then be separated from one another, as well, and used as a lightweight form of authentication.

That then explains the “wearable” part of “Wearables as Context for Guiard-abiding Bimanual Touch,” the title of my most recent paper, but what the heck does “Guiard-abiding” mean?

Well, this is a reference to classic work by a research colleague, Yves Guiard, who is famous for a 1987 paper in which he made a number of key observations regarding how people use their hands — both of them — in everyday manual tasks.

Particularly, in a skilled manipulative task such as writing on a piece of paper, Yves pointed out (assuming a right-handed individual) three general principles:

  • Left hand precedence: The action of the left hand precedes the action of the right; the non-preferred hand first positions and orients the piece of paper, and only then does the pen (held in the preferred hand, of course) begin to write.
  • Differentiation in scale: The action of the left hand tends to occur at a larger temporal and spatial scale of motion; the positioning (and re-positioning) of the paper tends to be infrequent and relatively coarse compared to the high-frequency, precise motions of the pen in the preferred hand.
  • Right-to-Left Spatial Reference: The left hand sets a frame of reference for the action of the right; the left hand defines the position and orientation of the work-space into which the preferred hand inserts its contributions, in this example via the manipulation of a hand-held implement — the pen.

Well, as it turns out these three principles are very deep and general, and they can yield great insight into how to design interactions that fully take advantage of people’s everyday skills for two-handed (“bimanual”) manipulation — another aspect of “touch” that interaction designers have yet to fully leverage for natural interaction with computers.

This paper is a long way from a complete solution to the paucity of modern touch-screens but hopefully by pointing out the problem and illustrating some consequences of augmenting touch with additional context (whether provided through wearables or other means), this work can lead to more truly “natural” touch interaction — allowing for simultaneous interaction by multiple users, both of whom can make full and complementary use of their hard-won manual skill with both hands — in the near future.


Wearables (fitness band and ring) provide missing context (who touches, and with what hand) for direct-touch bimanual interactions.Andrew M. Webb, Michel Pahud, Ken Hinckley, and Bill Buxton. 2016. Wearables as Context for Guiard-abiding Bimanual Touch. In Proceedings of the 29th Annual ACM Symposium on User Interface Software and Technology (UIST ’16). ACM, New York, NY, USA, 287-300. Tokyo, Japan, Oct. 16-19, 2016. https://doi.org/10.1145/2984511.2984564
[PDF] [Talk slides PDF] [Full video – MP4] [Watch 30 second preview on YouTube]

Book Chapter: Inking Outside the Box — How Context Sensing Affords More Natural Pen (and Touch) Computing

“Pen” and “Touch” are terms that tend to be taken for granted these days in the context of interaction with mobiles, tablets, and electronic-whiteboards alike.

Yet, as I have discussed in many articles here, even in the simplest combination of these modalities — that of “Pen + Touch” — new opportunities for interaction design abound.

And from this perspective we can go much further still.

Take “touch,” for example.

What does this term really mean in the context of input to computers?

Is it just when the user intentionally moves a finger into contact with the screen?

What if the palm accidentally brushes the display instead — is that still “touch?”

Or how about the off-hand, which plays a critical but oft-unnoticed role in gripping and skillfully orienting the device for the action of the preferred hand? Isn’t that an important part of “touch” as well?

Well, there’s good reason to argue that from the human perspective, these are all “touch,” even though most existing devices only generate a touch-event at the moment when a finger comes into contact with the screen.

Clearly, this is a very limited view, and clearly with greater insight of the context surrounding a particular touch (or pen, or pen + touch) event, we could enhance the naturalness of working with computers considerably.

This chapter, then, works through a series of examples and perspectives which demonstrate how much richness there is in such a re-conception of direct interaction with computers, and thereby suggests some directions for future innovations and richer, far more expressive interactions.


Thumbnail - Inking Outside the Box book chapterHinckley, K., Buxton, B., Inking Outside the Box: How Context Sensing Affords More Natural Pen (and Touch) Computing. 2016. Appears as Chapter 3 in Revolutionizing Education with Digital Ink: The Impact of Pen and Touch Technology on Education (Human-Computer Interaction Series), First Edition (2016). Ed. by Tracy Hammond, Stephanie Valentine, & Aaron Adler. Published by Springer, June 13, 2016.  [PDF – Author’s Draft]

P.S.: I’ve linked to the draft of the chapter that I submitted to the publisher, rather than the final version, as the published copy-edit muddied the writing by a gross misapplication of the Chicago Manual of Style, and in so doing introduced many semantic errors as well. Despite my best efforts I was not able to convince the publisher to fully reverse these undesired and unfortunate “improvements.” As such, my draft may contain some typographical errors or other minor discrepancies from the published version, but it is the authoritative version as far as I am concerned.