WiiLi.org

[FlyinWhee]

I was wondering, given the data obtained from the Wiimote's movement; be it in GlovePIE or any other interfacing method for the PC, if, and how, it would be possible to determine the position (X,Y,Z) of the wiimote in a defined space.

As far as I know, all movement with the wiimote is relative. as such, it would be possible to read a position as "the center", and interpret all accelerometer data from that point on to move the cursor around that spot. It would probably require setting the boudaries of that spot the make the data useable in the least...

But my question really is this, how do I figure movement in each direction from the data of the wiimote? I was looking at the data in WiinRemote, mostly, and the movement of the Wiimote other than rotation is eluding me; I don't see what changes when I go along any specific axis.

Any help will be appreciated, semi-concrete examples will be appreciated (in GlovePIE if nothing else), thank you.

[squeakypants]

The problem with accelerometers are that they can only fully track motion in a 0-gravity environment or while in conjunction with gyroscopes. However, if someone developed an algorithm to separate motion-based acceleration from gravitational, it would be possible (it would also make rotation-calculating much more accurate)

[CarlKenner]

Currently GlovePIE attempts to seperate rotation as Pitch and Roll, from acceleration as RelAccX, RelAccY, and RelAccZ. It doesn't do a very good job though.

RelAccX, RelAccY and RelAccZ indicate the acceleration in m/s/s. You add the acceleration multiplied by the time since the last update to the velocity. Then you add the velocity multiplied by the time since the last update to the position.

Unfortunately I couldn't get it to calculate sensible velocities, let alone sensible positions. I'll have another go later.

[alecool]

I have made a small program using dark basic pro and glovepie, which sends the acceleration datas and IR positioning datas through PPjoy virtual joystick. I achieved to mahe a small sphere moves like the wiimote, but I have to keeps it in front of the sensor bar. I can use all of the datas given by the wiimote. I use the distance between the two IR points to get the depth of the wiimote, and the Y and X position to put these coordinates to the sphere. I can also use the pitch and roll to rotate the model. I can't give the program now cause i went back home for christmas, but I will make some improvements and show what is possible to do with this...
_________________
French wiimote-on-PC user, so excuse my poor language

[FlyinWhee]

Well I'll fiddle around with what Carl posted, see If I can myself make some headway on the topic.

My idea is to make a small virtual space with planes that trigger events when the Wiimote crosses it, with a set of conditions.

The first idea is to make a much more logical air drum than what is currently available, but the ways of using such a flexible interface could be many.

Anyway, back to "work"!

[CarlKenner]

[squeakypants]

I just got an amazing idea. I've been thinking "You can't get rid of the gravitational acceleration without knowing the rotation, so you need gyroscopes to work in conjunction to be able to know the non-gravitational acceleration." However, we have the IR sources! If you have 2 IR sources, you can easily find the roll, which alone factors out g-acceleration from the X and Y axes. However, this brings a problem in itself:

The g-acceleration is used to find rotation using the following formula (note, all following formulas disregard negatives, absolute value must be used in many cases):