Fitts Law, applying it to touch screens_问答_开发者

Been reading a lot into UI design lately and Fitt's Law keeps popping up.

Now from what I gather its basically the larger an item is, and the closer it is to your cursor, the easier it is to click on.

So what about touch screen devices where the input comes from multiple touches or just single touches.

What are the fundamentals to take into account considering this?

Should it be something like, the hands of the user are on the sides of the device so the buttons should be close to the left and right hand sides of the devic开发者_如何学编程e?

Thanks

I started thinking about this recently too, and here some considerations:

Fitts' law was developed in the 50's as a human factors model (read: controls for fighter plane cockpits) so seeing it re-applied to human motor skills is actually just coming full circle. It definitely applies to mobile devices. [Historical note: The finding that it applied to mouse interfaces was actually a big deal at the time.]
One thing to note is that the Fitts'-endowed advantages of the edges and especially corners of the screen no longer exist on a touch interface: the "infinite size" only applies to mousing interfaces since the cursor cannot move past the edges. Obviously, the same limitation does not exist for our fingers. Basically, the edges are no better than the middle of the screen except for the potentially shorter distance to the target.
Here 1 (pdf) is an '06 study about optimal target sizes for one-handed thumb use, taking into account freedom of movement and such. I was hoping to find a paper that would be able to provide a modification or a new constant to Fitts' law for accuracy of the touch interface but a cursory search didn't turn one up. I guess that means I found a potential research topic ;)
I think one general conclusion to be made based on application to Fitts' law to smaller-screened mobiles is that it's hard to make usable widget-based interfaces without seriously sacrificing information density. One interesting alternative is gesture-based interfaces (beyond the popular pinch and zoom). Unfortunately, the lack of popularity and conventions makes the learning curve rather high. Mobiles are definitely one place that it might be worth the trade-off, though. I predict wider adoption of gesture interfaces on mobiles once conventions stabilize.

Yes, for a touch screen Fitts' law has to be applied in three dimensions, so it's different from the classical mouse movement considerations.

As you say, the origin of the movement is often the default position of the finger. This varies a lot depending on the device where the screen is mounted. On a hand held device you might use the index finger of one hand, or the thumbs of both hands, depending on the design.

Also, on a touch screen you have to move the fingers away from the screen to see it, which makes the distance between controls less important as you move back to the default position between clicks.

What to consider besides Fitts' law is the intuitiveness of the interface. If a button appears where it's not expected, it doesn't matter how close it is, it will still take time to find it.

One specific idea that attempts to leverage Fitts law is to put the most often used controls at the bottom of the screen (i.e., the opposite of current GUI conventions with the menubar and toolbar). This allows users to touch multiple controls in sequence without withdrawing their hands to see the effects, shortening the mean distance moved between inputs. For a tablet, kiosk, or desktop device, the bottom of the screen is probably also the hands’ “rest” position. However, there is the potential problem of the most important controls being the last thing the users sees when scanning the display.

Fitt's Law "predicts that the time required to rapidly move to a target area is a function of the distance to and the size of the target." What's important isn't that Fitts discovered this (it is obvious) what he noticed was that the increase due to distance and size fit a logarithmic formula, which the law models.

On a Windows-Icon-Menu-Pointer (WIMP) system, what's important is that you have 1 location with zero distance (where the cursor is currently) and 4 locations that are of infinite size (the edges of the screen, which the pointer cannot extend beyond). That's really why fitts law pops up so much in UI design (aside from giving weight to things like "Don't make tiny buttons", etc)

But the law makes a lot of assumptions about the range of motion you have available with your hands. If you're holding a tablet with two hands, the law goes out the window. If you're holding it with your left hand, then things on the right side will be easier to reach, etc. So its going to be a lot harder to make generalizations than with a pointer.

That said:

Think about where the users hands are going to be, and if they're both going to be free or not. Place buttons closest to where you think hands will be.
Cluster buttons such that you aren't requiring the user to make a variety of successive taps that are far apart (unless, of course, you're designing a game, in which case that's part of the skill)

Well, you should design for the most important fingers, after all (index, for example). Not that you shouldn't use the others, of course, but people generally are more geared toward using some fingers in detriment of others.

I don't think you can provide a general answer that will work across all sizes and types of touch screens. For example: The infrared vision technology on the Microsoft Surface can fail if a user has extremely dark finger tips (very very rare), but this would not be an issue on a capacitance based touchscreen.

The best practice to implement is lots of testing, with a variety of users. You will quickly learn what works on you device and what doesn't.

I did a paper on this for my graduate human computer interaction class using evolutionary computation to design a more efficient keyboard based on the domain of text that was being typed. I really should release it as an iphone/droid app.