Detect direction of motion

Yes (and obviously so, since you included "anything else" in the list of possible options ;-). But it'll all depend on whether you want to measure motion in a closed room, or out in the open. Without walls to reflect sound or light off, or some other feature that can be observed to change, it's essentially impossible to measure movement.

But does any of this have to do with algorithms of computer graphics?

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

If you need only direction after a change in direction, use a 2D accelerometer.

--
------------------------------------------------------------
Creepy, Soulless Gigolo for President ? NOT !
---------------------------------------------
THK is one weird, weird something.

Yes.

Oh you want a more detailed answer, without more details especially about the object, distances, response time, frequency of results, and environment no more details can be given.

--
Paul Carpenter		| paul@pcserv.demon.co.uk
        Main Site
              GNU H8 & mailing list info.
             For those web sites you hate.

The device could be used outdoor or indoor.

But not totally dark envirnment

That doesn't give the necessary information. Let me show you why: imagine your device is were to use an optical pattern-recognition (rather than signal echo timing) based method. If you try to use that device outdoors, and 'outdoors' just happens to be right in the middle of one of the salt seas of Utah, it'll fail completely, because the environment will look exactly the same, no matter how fast it moves.

Ultimately, you need something external to the object that specifies what "being at rest" means to your device, because physics doesn't provide you with any such thing. That's what Einstein's Principle of Relativity really says: there's no such thing as absolute speed (and thus no way to measure it, either). So you have to measure speed relative to *something*, and the device must be able to determine where this "something" is.

Optical pattern recognition would work in some cases (like in the optical mouse you may have attached to your computer), ultrasound can work, too (or bats wouldn't manage to catch anything at night).

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

How about a 3D inertial measurement sensor? Not cheap but it should do the job..

Rob

Is there any way to focus those little optical mouse sensors on infinity (or a couple of feet on out to infinity)? I would think those would work well against a fixed background, and they're cheap.

-----

No. This does not work when light is insufficient.

The more complicated solution would be accelerometers and laser gyroscopes to computer distance moved. Bloody complex if u ask me though I've never done it myself.

What is the theory behind, using what sensors?

Not just more complicated, but probably unusable unless I misread the original question. An accelerometer can't actually measure velocity

--- if it could, it'd be called a velocimeter. You can deduce speed from accelerometer readings, sure, but only if you add

1) highly accurate summation of acceleration over time 2) regular calibration to tell the device what its current speed really is, to serve as starting value for the integrator.

IOW: without a separate method to measure speeds, an accelerometer can't measure speed at all (nor position, for that matter). A classical chicken-and-egg situation.

--
Hans-Bernhard Broeker (broeker@physik.rwth-aachen.de)
Even if all the snow were burnt, ashes would remain.

Unless I misread it, it asked only about direction.

--
------------------------------------------------------------
Creepy, Soulless Gigolo for President ? NOT !
---------------------------------------------
THK is one weird, weird something.

It asked about direction of *motion*. Not direction of acceleration. An accelerometer can only measure the magnitude and direction of the _acceleration_ (delta v) vector.

You have to integrate acceleration over time to get velocity, and you have to know what the starting value of velocity was.

That's why my response from 3 days ago read:

If you need only direction after a change in direction, use a 2D accelerometer.

If you model velocity as a vector, a change in direction will register an acceleration (delta v or force) without a change in magnitude (velocity). This requires a little understanding of Newton's 1st and 2nd laws of motion. It's why you get pushed sideways when driving around a radius at constant linear velocity.

If you want to pop in at an arbitrary slot in time and determine direction without any history, that is maybe the problem at hand. A poorly stated problem results in many interesting threads sometimes, where a consisely stated problem causes useful and terse responses. If you want to increase the volume of C.A.E., encourage poorly stated questions.

--
------------------------------------------------------------
Creepy, Soulless Gigolo for President ? NOT !
---------------------------------------------
THK is one weird, weird something.