Tangible Interfaces for Computers

Jesrad writes "A friend pointed me to this impressive demonstration of the SenseTable by James Patten, of the Tangible Media Group project of the MIT. This project aims at conceiving better human-machine interfaces by using the concept of physical objects that the user can manipulate, to represent abstract computer data and commands. The device looks and works a lot like what was envisioned in Minority Report, it uses pressure to track blocks on a sensitive surface, and feeds back to the user by superimposing graphical data. Want to change the volume of your MP3 player? Just put a block on it and turn like you would a radio knob. Menus and commands are accessed by moving a block along command hierarchy, represented in a simple tree, or by touching the command's name. So far it only lacks a device for text input, like a keyboard, but maybe voice recognition will replace it?"

Keyboard implementation should be easy...

[internet-redstar]

... just press the button to type the specific character?
One could even have different keyboard layouts being switchable with a knob... oh, wonder, wonder!

Feel free to add other irony below...

umm...

[mozumder]

So why can't you just put a volume knob on that MP3 player?

Oral audible hell

[kfg]

"So far it only lacks a device for text input, like a keyboard, but maybe voice recognition will replace it?"

Or maybe they'll just plug a keyboard into it? Voice recognition may well have its uses, especially as an accessibility technology, but as a general input device it's really a pretty poor idea.

Unless we're all supposed to sit in a cone of silence or something.

KFG

Call me a skeptic

[GFW]

While various varieties of tangible interfaces might be useful in specific circumstances, the typical user doesn't want more crap on their desk. They want a flat, easily positioned, brillant screen (or three). They want a keyboard (which could be virtual, but most people prefer some tactile feedback for typing). They want something for pointing (which could be a glove, a mouse, entirely virtual, ...) They don't want a metaphor that looks like Play-School.

Re:Call me a skeptic

[Anonymous Coward]

Additionally, it doesn't seem to me that there's much of a difference between this and current user interfaces. It looks shiny, but basically what they do is use a block of wood to point at things instead of a mouse. And orientation matters as well as position. Other than that, it's just drag-and-drop and point-and-click, except without mouse buttons and with shiny lights...

Lourens

Voice recognition vs. keyboards

[LeoDV]

Even when the technology is perfected to Star Trek standards i.e. you don't even need to think about articulating to make yourself understood by the computer, keyboards will remain the preferred input method of many, including me, simply because it's the fastest. I haven't ever "learned" to type but I average around 100 WPM and peak at 120, without a DVORAK keyboard. I'll rather use that to jolt down an idea, write a letter, program or post at Slashdot than voice recognition.

Defense of tangible interfaces

[G4from128k]

I can understand why some people are appalled by tangible interface concepts. These are the same people that refered to GUIs as WIMPs (Windows, Icons, Menus, and Pointers). For some people, a command line, keyboard-coded interface just works. But it is not the best interface for everyone or every application.

1) Media creation: Who still creates CAD drawings with a keyboard only? I used some early versions of Autocad that where keyboard-only -- they sucked. Sometimes a tangible pointer with a 1-to-1 interface mapping between a 2-D surface and the screen is superior. For artists, the use of an LCD graphics pad and pressure-sensitive stylus means much higher productivity and finer control. (I've even scene academic research suggesting that a two-mouse interface could improve productivity.)

2) Mapping to the Realworld: Go aboard an aircraft carrier and look at how they keep track of flight-deck operations. A miniature replica of the flight deck and miniature aircraft provide an intuitive 1-to-1 mapping between the model and the real-world. I'd bet that they could improve flight deck operations if those little aircraft moved automatically to reflect actual locations and if manual movements of aircraft spawned automatic commands to flight deck personel.

3) Multiuser interfaces: the demos of MIT's system that I have seen (a business-oriented supply chain visualization tool) leverage the table interface for multi-user applications. With the table, anyone around it can reach over and move a block. And everyone can easily see who moved the block.

The power of tangible interfaces is that they can help create a more literal mapping between a digital artifact and the real-world. Sometimes less abstraction leads to better ease-of-use.

Fascination with voice recognition, what gives?

[Doomdark]

So far it only lacks a device for text input, like a keyboard, but maybe voice recognition will replace it?"

I'm certainly not the first poster to comment on this, but I just don't understand why many assume voice input would be the preferred method? That it'd even be better than physical controls (be that keyboard, mouse, switches, joystick, whatever). There's enough aural noise in the environment, even without more; accidental commands, specificity, technical things... And except for niches where it does make sense (if one can not use his/her hands or even legs), there just doesn't seem to be much beyond 'coolness factor'? Just like you can get carpal tunnel syndrome, your throat can go sore etc.; there are no health benefits; people can generally point/click/type faster than talk; GUIs are multi-dimensional (2 currently), speech generally single-dimensional, so there's one less way to distinguish what was the target (ie. no location information)... and so on and on.

Now as to Star Trek and other sci-fi movies (including Minority Report), isn't it fairly obvious why voice input was/is used? It's the easy way to indicate what a character is inputting, and what are the results! Even if it wasn't for futuristic touch, it's so much better for needs of movies and tv series than, say, keyboard input (keyboard and mouse are only shown when realism is needed). Directors are in general experienced and smart professionals, and know that voice input is a very good solution for THEIR problems. Just like even though there hasn't been the need to stay on call for tracing to work for decades now, they still always imply it is, in crime series, just because that's a cheap (albeit unrealistic) trick to add some suspense to plot. Just don't assume they are prophets that show what future will be; just what works for them.

Re:Call me a skeptic

[gidds]

How many people said something similar when the WIMP environment (e.g. Mac) went public? "Real computers need you to type everything! Anything worthwhile can be shown as text - if I want to see pretty pictures I'll go to an art gallery! And keep those mice in the toybox where they belong!"

Initially, that took lots of space, seemed a waste of resources, and you couldn't do much with it. Since then, resources have increased tremendously, new applications and methods have been developed that make good use of it, and people see the extra desk space as worthwhile. I don't know if the same might happen to the SenseTable, but I do know that if so, it won't be because it fits today's hardware, apps, and interfaces, but because it'll fit tomorrow's.

We've seen this before...

[brain159]

At a glance, this sounds very much like the underlying interface stuff behind the Audiopad project (also from MIT, IIRC) - smart pucks moved around a projected image on the sensing surface. There were a few pucks which controlled various musical loops and one which acted as the microphone (the closer a loop was to the mic puck, the louder it played).

Not that I'm doing anything down - my guess is they're now making more general use of the stuff they'd developed for Audiopad, or Audiopad was just the first application they'd come up with, or something along those lines. Nice to see the technology back again actually, I've got a video of Audiopad and it's pretty cool.

Beware geeks designing interfaces..

[adrianbaugh]

Just look what they did to emacs :-O
Seriously, while this probably has niche applications (previous posters have mentioned a few that sound plausible) I don't see that it offers much to the conventional desktop user (a keyboard and mouse require much less movement than the shenanigans Tom Cruise got up to in the movie and, other than keeping office workers fit, these interfaces will just lower productivity).
So what about wearable computers? Something you wear on your belt with a head-mounted display, designed to be used while walking along? Well, to me it doesn't make much sense in this context either: again, if you end up requiring much odd movement on the user's part it won't work. In my opinion the future is far more likely to look like a next-generation of Canon's eye-controlled (pupil-tracking) autofocus system to control a pointer on some head-mounted display coupled with (in the short term) an interface that minimizes the need for text input together with some kind of finger-based character input device[0] or (longer term) speech recognition of a standard where the software doesn't need training and can cope with background noise[1].

[0] There was one mentioned on slashdot ages ago that looked a bit like a gripmaster (key for each finger plus the thumb), and text was typed by entering chords.
[1] Incidentally, how much research has been done on using stereo input to speech recognition programs to reduce background noise? I would have thought that would help quite a lot, albeit at the expense of CPU time.

Re:Or, alternatively,

[netsrek]

Remember that the only intuitive interface is the nipple. Everything else is learned.

Stop repeating this crap. Have you ever watched a baby have to learn how to breast feed? There's a reflex there to get to the nipple, but actually doing the feeding isn't intuitive at all.

There are no intuitive interfaces, only ones which are similar to other interfaces you've already learnt how to use...