How does the Wii pointer work?

[thomase]

From what I understand, the wiimote itself has 3-axis tilt sensors that determine yaw, pitch, and roll over the full 360 degree range. The sensor bar will then be used to determine position in 3D space. A calibration process will ask you to point at specific locations on screen, and from there, the Wii can map any combination of 3D position and tilt to a point on screen. The sensor bar will not be needed for games that don't use pointing control.

The thing I'm missing is how its possibile to triangulate position in 3D space with an array of sensors arranged on a line (the sensor bar) rather than a plane. I understand that the sensor bar is not literally a line (i.e it has thickness), but it does seem rather thin...

 

[Zaphod]

No triangulation, methinks. The controller doesn't know exactly where it is in 3D space, but based on accelerometer data it's trivial for it to approximate where it is relative to where it was when it was 'calibrated' at the start of the game.

 

[thomase]

Some googling yielded this explanation:

"Sensor bar contains IR LEDs that flash in a predictable pattern, wiimote has a IR camera behind the window (think a cameraphone-grade camera). From this they can get orientation relative to the TV, and distance (by how wide the "sensor bar" is). Combined with the accelerometer-based orientation they can recover a full 3D position."

Even if the camera has a sampling rate high enough to resolve skew between between the IR LEDs and translate that into position, how can you get elevation from that?

 

[Shifty Geezer]

Is that fact or speculation? If true, using a 2D sensor, they wouldn't need two transmitters on the sensor bar. Also the sensor bar is a misnomer because it hasn't any sensors, only 'lights'. I'd be more inclined to think the Wiimote transmits an IR signal picked up by the sensor bar. A link would be appreciated.

 

[Bohdy]

The best explanation I have heard is that the technology uses a three-pronged attack:

1.) Miniature gyroscope within the controller to determine tilt (roll, pitch, yaw)
2.) 3-axis accelerometer to measure acceleration in 3D-Space (PS3's DS3 has the same capabilities up to this stage)
3.) Synchrony with the sensor bar (via IR) to ensure correct (relative!) placement detection in 3D space, necessary to offset the innacuracies from just using accelerometer data.

The "nunchuck" just uses an accelerometer in comparison, which should provide rough gesture detection.

And though it is concievable that the position and dimensions of the TV screen could be calibrated by pointing at the diagonal corners, which would allow it to function as a LoS pointer or lightgun, this has not been attempted by any developer to date. I think that sort of thing will be mostly avoided, as it adds more complexity to the experience, and relative control seems to be quite effective without absolute placement.

Edit: This thread on neogaf has some good information: http://www.neogaf.com/forum/showthread.php?t=99587

 

[Titanio]

Anandtech have a little more info on it, gleaned from a developer at E3:

http://www.anandtech.com/tradeshows/showdoc.aspx?i=2757

The sensor bar emits an infrared field out directly in front of the TV. In order to accomplish this, it is necessary to place the sensor bar in the same plane as the TV, either on top of or below the TV is recommended. As the remote is pointed towards the TV, it interacts with the infrared field. Using triangulation logic, the remote is able to determine location, angle, and distance; as you move the remote around, the change in location/angle/distance is calculated. The remote then communicates with the console via Bluetooth, sending this information to it, and from there the software determines how your movements relate to the game world/application.

The sensor bar doesn't receive any type of signal from the wireless remote and is there purely to generate the field. The remote also has a gyrometer and accelerometer to allow it, for instance in the tennis game, to know the difference between a back hand or regular swing. You need to maintain a proper distance from the sensor bar as well; if you come to within about three feet or less, it will not work precisely.

 

[Shifty Geezer]

So the sensor bar is a total misnomer. It's an emitter. I'm curious how it works though for location, angle and distance, if there's only one sensor at the end of the Wiimote. Also it surely can't detect tilt and roll which are the responsibilty of the MEMS.

 

[Squeak]

Even if the camera has a sampling rate high enough to resolve skew between between the IR LEDs and translate that into position, how can you get elevation from that?

(I take it you mean z-position?) By looking at the explosion/implosion of the two LEDs.

 

[Ty]

IR emitter? Sweet, I'll be able to use the Nightshot mode on my video camera with greater effect!

 

[LunchBox]

how big is the sensor bar?

 

[Arwin]

how big is the sensor bar?

Looks to be roughly about 1.5-2.0 the length of the wii-mote, and about half its width and same height. There's a good picture here, about 8 shots down.

http://www.engadget.com/2006/05/10/nintendos-e3-booth-tour/

 

[Shifty Geezer]

About 10" long?

 

[NANOTEC]

Looks to be roughly about 1.5-2.0 the length of the wii-mote, and about half its width and same height. There's a good picture here, about 8 shots down.

http://www.engadget.com/2006/05/10/nintendos-e3-booth-tour/

Whoever designed Nintendo's booth did an amazing job. It's so beautiful and futuristic.

 

[LunchBox]

Looks to be roughly about 1.5-2.0 the length of the wii-mote, and about half its width and same height. There's a good picture here, about 8 shots down.

http://www.engadget.com/2006/05/10/nintendos-e3-booth-tour/

About 10" long?

Ah I see...

thanks!

At least the bar is quite sleek...


On a side note, I like that dual stick controller that they have!!!!
http://www.engadget.com/media/2006/05/IMG_7941.jpg