Wii uses 3-axis acceleration sensors of STMicroelectronics for its controller

The rumble of the wiimote was said to be very weak by some E3-attending blogger, so any interference would also be weak. Why not simply have a highpass filter cut out all the tiny movements? (Or is that a lowpass filter? I always confuse the two. )

There might also be some kind of sensor in the rumbler so the Wii know how quickly the motor is turning, knowing the mass of the off-axis weight, that could also help eliminating unwanted noise.

I had hoped there would be a rumbler in the nunchaku thingy as well, seems there isn't.

 

First of all, the problem with rumbling is rather the amplitude (thus the intensity) than the frequency. Low pass lets low frequency pass, High pass lets high frequencies pass.

If the intensity, direction and the time profile of the rumbling is known, it can be compensated as long its within certain boundaries.

 

Well, sony seems to have decided their controller's acceleration controls was getting interference from rumble, and pulled rumble.

Is it any reason to belive vii is more "rough" on its motion detection than ps3, since ps3 dont got rumble?

 

No, not really. Motion sensors are motion sensors. Any motion sensor that can't pick up a vibrating weight from rumble is going to be too course to be any use for delicate control like Wii, I'd imagine.

BTW : Are Nintendo licensing from Immersion? I thought they avoided that ruckus in GC.

 

That would be lowpass. But the trouble is, where do you cut off? If set too low, it'll interfere with the actual movement of the controller.

That would be too complicated and thus too expensive, I doubt they'll do anything like that.

EDIT: since the speed is calculated as the integral of acceleration, the algorithm itself will probably have a slight low-pass effect. That's also where any additional "real" filtering will be done, I guess. At least that's how I did it in one of my old projects, worked like a charm for removing noise (though I had a very aggressive low-pass and that wouldn't work for this controller since it would introduce huge lag)

 

Either the "wiimote" filters out the rumble from the motion sensing data as suggested, the rumble is weak enough not to interfere too much with the motion sensing,
or the gyro based motion sensing just isn't functioning while there is a rumble effect (note, the IR based motion sensing is not affected by rumble).

But, the correct answer I guess is that as most of the motion sensing that needs accuracy is accomplished via the infrared sensor bar, the rumble has little effect on accuracy due to the broader movement of the "wiimote" than it would if the gyro was used for "cursor movement" and such.

 

Human hands are actually pretty inprecise. We can't do relatively fast, very small movements like a rumble shaker would. We can do either, just not both at the same time.

There's a double set of optodetectors in any ball mouse, and typically in any mouse with a scroller wheel regardless of if it's a ball mouse or not. I seriously doubt these components are particulary expensive. (I'm thinking fractions of a penny here in large quantities.)

Not that I really expect Nintendo to actually do this, I just mentioned it as a possibility...

 

Trembling hands would be a good example for stuff that has to be filtered out. Just any "small" random movements. These sensors are very precise, you can detect <1mm movements, that's why it always has to be filtered somehow.

The other issue is expensive because you'd have to exatly measure the rumbler's movements and compensate for these in realtime, which would be a bit too much for the controller which costs a few bucks. In itself, it's of course possible to do it.

 

Nope, they've licensed their own rumbling patent. Their technology is actually different from Immersion's patents.

 

That's also my guess. I'd like to know more about how the sensor bar affects the system. From some gamespot editors, I got the impression that it plays a very important role, to the point where if the controller is too close or too high, hardly any sensory input is registered at all. So does that mean most of the input actually comes from the sensor bar, and relatively little from the other devices?

This would then be rather different from the PS3, where all movement is detected within the controller.

 

It sounds to me like the sensors in the bar are quite directional. Or the Wiimote's transmission is. That could suggest an implementation measuring signal strength from the Wiimote at the two sensors and determining it's horizontal positon. Dunno how that'd get vertical postion though

 

AFAIK the IR is for the cursor/pointer 2D. The tilt sensors operate independently from the IR sensor which track 3D movement.

 

I think the trouble with the flittering of the cursor in many of those videos had little to do with lack of filtering and more to do with people standing up when trying it for the first time. Try using a mouse when standing, in a stretched out arm.
It will get much better when people are sitting down with their arm rested in the lap.

 

That would make sense and explain it maybe partly. Movement created by vibration is typically exclusively x,y,z, and doesn't really cause any tilting to happen. If the motion was detected in the controller, that could greatly affect registered movement, but relative movement of the controller vis-a-vis the sensor bar would be very small and probably not even need to be compensated as it would either not register in the first place ('deadzone') or be very slight, at best adding a subtle visual cue - but no violent shaking that would be registered if the motion sensor were on board (and thus very different in nature from the IR sensor.

That also means that movement detection will be the characteristic and defining difference between the PS3 controller and the wiimote. I can see advantages and disadvantages to both.

 

I was just wondering if the sensors might be seperated inside and contained in a sort of "dampening chamber or encloser" for the sensors that would help limit the vibration and negative effects to the sensors. So you are feeling the rumble but vibration has limited effect to the sensors (and tilt and movement would still be detected and maintained). This along with some filtering code to help any residual effects if needed. (Thinking of how rocket gyros are contained in a dampening encloser to help prevent the vibration from the engines etc. which has negative impacts on their data.)

All this is just a theory on my part

 

jpr27: nice theory, but not really practicable besides being redundant (the SW can do more than anything mechanical)

I guess they'll just filter it out through SW since that won't affect the precision/sensitivity of the controller that much thanks to a rather weak signal. I suppose you don't need very high sensitivity for the controller.

 

It might not be as bad as I had initially thought. Assuming that you can limit the rotating mass' motion to a plane (let's say x-y), you should see a nice sine wave in your x-direction signal and the same in the y-direction, delayed by 90 degrees. You know how fast the motor is turning, so just filter at that frequency.

The sensor is actually an accelerometer, not a motion sensor. Not much motion is generated by a rumble motor, but the acceleration is probably a lot more. This, along with running off of batteries, may be why the rumble feature is so weak.

 

I doubt the Wiimote filters anything at all. So in the strictest sense, tiny vibrations will affect motion. However, the user interface is nowhere near this precise.

Think about it. If minute vibrations could impact motion, then movement would be erratic and stability a Herculean task for humans and robots alike.

But with a dumbed down interface, small amounts of energy -- be it electrical, mechanical or whatever -- can affect the system without consequence.

 

From what I've read, all pointer style play is done using IR and the sensor bar. The tilt functionality is achieved using the accelerometer. If the rumble motor is turning at 10 Hz, how is the hardware to tell that you're not fluttering the controller back and forth at that frequency?

 

With Bluetooth, Wiimote doesn't have to be aimed at anything. Sensory feedback updates the console of its orientation and animation. So you can setup Wii over there and play in front of a screen over here (as they did at E3) without having to point it.

The sampling rate or frequency would have to be much lower. So instead of gathering input data at 10Hz or more, the hardware might look for it, oh I dunno, once or twice a second.