Next Generation of Gyroscopic Controllers on the Horizon

Jamie found a story about a next gen input device that is functionally similiar to the Wii, but instead of using IR, it gets all location information from gyroscopes and accelerometers. This has the potential to be more accurate and maybe not require me to contort my wrist to bizarre angles in order to successfully collect the stars that are like oxygen to me.

Comments

This is the Wii remote, minus functionality

[hansamurai]

it gets all location information from gyroscopes and accelerometers. This is exactly what the Wii remote is, except the Wii remote adds in the functionality of the IR pointing device. The Wii remote is built on two ideas: the gyroscopes and accelerometers delivering feedback on movements and the IR device allowing it to interact directly with the television. What Jamie so cleverly found is a device that is only the first half of what the Wii remote is built on.

Re:This is the Wii remote, minus functionality

[Gravatron]

wiimotes don't have gyros in them, IIRC. Just an accelerameter.

Re:This is the Wii remote, minus functionality

[Kaeluka]

I worked on an application using quite similar accelerometers, a few years ago. We used the ADXL103, while the Wiimote uses the ADXL303, which is basically the same sensor, but with three measured directions. What we tried to do was measuring distances - quite similar to the Wiimotes goals. What we found out, however, was, that the sensor just isn't exact enough to provide reliable velocity- or even position-Information. The problem is, that the provided accelaration-data have to be integrated (even twice, for position-measuring). That can't be done so easy, without adding up the noise as well. It wasn't too much of a problem in the end, because we needed the sensors for demo-purposes only, but that was the problem we found out. Here is a link to a picture. We used three sensors to measure X, Y and Z-directions, the lines should return to zero after the vibrations, but they didn't ;) http://img255.imageshack.us/img255/85/komischacropfz8.jpg I don't say, that it can't be done, what I'm saying is: it might be hard.

Re:This is the Wii remote, minus functionality

[randyest]

YDNRC; neither the nunchuck nor the wiimote contains any gyroscopes. Both, however, contain accelerometers, and the wiimote adds an infrared camera (which "sees" the sensor bar IR LEDs) to correct drift error. You might note that the page you linked does not contain the word gyroscope.

"collect the stars"?

[Speare]

maybe not require me to contort my wrist to bizarre angles in order to successfully collect the stars that are like oxygen to me

If that was a reference to Super Mario Galaxy, I'd have to say you must be playing wrong. SMG leverages far less Wii controller range of motion than most other Wii games I've tried. WarioWare Smooth Moves gives a bunch of cute names to various Wiimote controller positions, so it's handy to talk about other games with these terms too. SMG just uses "Remote Control" and "Umbrella" postures, and to spin you need to shake the Wiimote a little. If you want wacky untenable wrist positions, try some of the later levels of Kokorinpa (Marble Mania). There are wrist positions in that game that even Smooth Moves didn't try to name, but I'll call them "Policeman's Thumblock" and "Say Uncle."

Wii already uses MEMS accelerometers

[Sockatume]

As the comment title states. The acceleration due to gravity from the Earth allows it to track which way is down, too, avoiding the need for little spinning gyroscopes. What did submitter think the Wii used to track movement when the remote wasn't pointed at the IR sensor bar? Psychic powers?

Re:Wii already uses MEMS accelerometers

[grumbel]

I find it highly unlikely that the Wii controllers are going to use that information in any meaningful way. All the Wii games that require tilt information (i.e. almost all) use gravity to get it. It is really pretty simple, the Wiimote returns acceleration in X, Y and Z direction, whenever the acceleration is not null it is gravity accelerating the Wiimote. (x,y,z) simply becomes the vector that points down to earth, which allows you to calculate pitch and roll rather easily, but not yaw, Wiimote can't track that. This of course only works as long as you hold the Wiimote still, as soon as you move it around the Wiimote can't really track anything, which is why you see so many waggle in games and no real 1:1 mapping.

what happens if I'm holding the controller (as opposed to dropping it and letting gravity "accelerate" it downward)? When you drop the controller the accelerometer doesn't register any acceleration, its when you hold it still that it does. One of the joys of physics that might not be obvious at first.

Article not very accurate...

[sirwired]

The Wii Remote tracks its position via an infrared sensor that users must attach to their televisions.

Firstly, while it is called the "sensor bar", it isn't a sensor at all, it is just a row of IR emitters. There is no receiver on the bar. Instead, there is an infrared camera in the Wii Remote that takes a "picture" of the bar to figure out which way the remote is pointed.

Also, the IR system is only used as as calibration for the accelerometers. The accelerometers in the Wii Remote still do the bulk of the work. If the Wii Remote relied on the IR camera as the primary sensor, it would be useless every time line of sight to the sensor bar was lost. What the Wii Remote does is keep rough track of remote position using the accelerometers, and then when the camera is pointed at the sensor bar, it re-calculates the starting point for the motion tracking to start from.

As far as this outfit using the fact that golf on the Wii leads to bad golf habits in real life: Duh! The Wii is a toy; it is not meant to be an accurate golfing simulator.

I can fully understand Nintendo not putting gyro's in the Wii Remote. It would have driven up the cost, reduced battery life, and introduced a moving part just begging to break.

SirWired

Re:Article not very accurate...

[Paralizer]

Also, the IR system is only used as as calibration for the accelerometers. The accelerometers in the Wii Remote still do the bulk of the work. If the Wii Remote relied on the IR camera as the primary sensor, it would be useless every time line of sight to the sensor bar was lost. What the Wii Remote does is keep rough track of remote position using the accelerometers, and then when the camera is pointed at the sensor bar, it re-calculates the starting point for the motion tracking to start from. The wiimote has factory set calibrations, it doesn't recalibrate itself on-the-fly. The only thing IR is used for is limited yaw calculations, depth (distance from the sensor bar) estimation, and calculating the X,Y position you are pointing at on the screen. For these the sensor bar is used as a primary sensor because that's the only point of reference that can be used to get that information. Everything else you said is spot on though.

Re:Article not very accurate...

[BitZtream]

I guess you haven't seen the stuff done by Jason Lee ... the accelerometers do not have anything to do with where your cursor is on the screen, it is all the IR.

This is why he has a Wii remote held stationary, with infrared emitters being used in place of the sensor bar to get various neat user interfaces to work.

Accelerometers are susceptible to cumulative errors that would make them useless for an accurate pointing device. The accelerometers in the Wii remote help it detect the angle its at (wii sports uses for swinging the club/racket/bat/bowling throws), and vibration/shock (when you wiggle it in Zelda or Mario Galaxy). Combining angle and acceleration data (by working out what the acceleration is after gravity has been removed from the sum). Unless you can sample at an infinite rate, with infinite precision, you can't avoid the errors. And frankly, why bother, the IR sensor requires less processing power and is more accurate since its position (in theory) in a way that aligns it with the device you're supposed to be pointing at in the first place.

The accelerometers can be calibrated without any reference to the sensor bar by simply holding the Wii remote still. You can assume the Wii remote is only limited in the distance it will move in any given period of time since in the context of the system, its only going to be able move a few feet, the range of your arm swing. So, given that they know its never going to be in constant motion, and at some point it has to stop, or its moving in a circle, you can, once you have a reading across all the accelerometers that measures close to the force of gravity and no more or less, if that continues for more than a few seconds, its likely the device isn't moving and calibration can occur. It really only needs to calibrate with the sensor bar once to get a baseline for how much the accelerometers are off. Since the accelerometers aren't going to be changing temperature drastically even in some players sweaty hands, you hardly have to even consider adjusting for tempature drift after the first calibration. They probably do just because its likely the accelorometers used already have temperature sensors on them for just that purpose anyway. So ... finding 'down' at any given point in time is relatively easy, using that you know the angle of the Wii remote at any given point in time. It doesn't help you at all as far as which direction the remote is pointing (north/east/west/south) but its perfect for that golf swing or driving game where you tilt it for steering.

As a pointing device, the Wii remote IS useless every time it loses site of the sensor bar. As anyone who owns a Wii can confirm, its a rather annoying fact of using the Wii remote.

'Gyros' no longer require moving parts. Technically, REAL gyros do, but no one uses them anymore if they can avoid it due to power consumption and reliability due to moving parts and wear. Now 'gyros' for sensing rotation around an axis use piezo films which detect based on how the film bends like modern accelerometers (and as such are skewed by gravity) or MEMS technology which detect by shooting a laser into two strands of fiber optic material around a circular path and measure the time difference between when the beam arrives back through each path to detect rotation without being skewed by other accelerations/gravity. They are used in many guidance systems to supplement GPS data for more accuracy by helping to correct GPS measurements between GPS updates. Or, in purely inertial guidance systems. Of course, just like accelerometers, they are susceptible to cumulative errors as time goes on. With a combination of Gyros and accelerometers and our good friend gravity, a large amount of the errors can be corrected for, but not completely.

Here are some examples of MEMS gyros from Analog Devices: http://www.analog.com/en/subCat/0,2879,764%255F801%255F0%255F%255F0%255F,00.html

Re:Article not very accurate...

[grumbel]

The Wiimote, like any device that tried to figure out position and velocity from an acceleromete You assume that the Wii games do that, but they don't. There aren't games that care about the Wiimotes position in space, they care about acceleration and orientation and little else. The Wiimote simple doesn't have sensors to give position tracking, for one thing the sensorbar only gives you 2 point positions, from which you can derive distance and rotation relative to the sensorbar, but little else, so it is useless for calibration purpose. But more importantly, trying to track position with an accelerometer alone would never work. An accelerometer could only do that if you hold the controller steady, as soon you rotate it around all the accelerometer data gets messed up with gravity and you can no longer know what is gravity and what is actual movement. It also becomes impossible to properly figure out the orientation of the controller. So any try to get position data would get messed up the very second the user moves the Wiimote, which would make it a very pointless exercise to begin with.

Re:I'll stick with the mouse...

[ezzzD55J]

http://www.hacker-dictionary.com/terms/gorilla-arm :)

IR has more than Gyroscopes

[Paralizer]

Motion detection and pointing are really two different features.

The wiimote uses a 3-axis accelerometer to calculate roll, yaw, as well as gravity forces on each axis; this gives everything except for yaw. Yaw would be nice but it really isn't needed because the shape of the object in your hand doesn't feel nature (or comfortable) to rotate it in that direction. To solve the issue of pointing the wiimote uses a camera sensitive to IR light and captures it at 1024x768 resolution. Since the wii sensor bar has two dots the virtual screen resolution is actually slightly bigger than that.

In addition to actually pointing at the screen the wiimote's IR camera can also be used to estimate the distance you are from the TV. So in all the wiimote has several degrees of freedom:

• Pitch
• Roll
• Yaw (very limited with IR)
• Distance
• X,Y position on the TV

With all of this you get a pretty decent idea of where the wiimote is in 3d space and at what orientation it is at.

Now consider the distance you are from your TV. As you move further away then the angle you should move the wiimote either up, down, left, or right should also decrease because the object. The wiimote can do that because of the IR camera. If you use a gyroscope you lose this because no matter where you are in 3d space it only cares about the orientation it is with respect to gravity. So if you are aiming at the top of your TV and you move backwards with a gyroscope then it will still be aiming at the top, whereas with the wiimote it will go above the top because that's where you're actually pointing the device.

With a gyroscope and accelerometer you would get:

• Pitch
• Roll
• Yaw

That's about it. No distance, no X,Y position.

I think the wiimote still wins out. The only thing that I would change with the wiimote is give it a higher resolution IR camera, but maybe that was too expensive for Nintendo (that may also have been a reason they didn't do high def?)

Also you can already use the wiimote on a PC for free and have millions of potential customers already owning one. So why would anyone want to pay royalties to use this thing?

Re:I'll stick with the mouse...

[MobileTatsu-NJG]

Any input device that requires you to continually keep your hands elevated will never work. Not to mention, constant movement.

The reason a mouse and keyboard is so effective is because you can use them both all day long with little to no effort. That's how the Wii got it right. It's like using a mouse with gesture controls. It's not 1:1 movement like people dream of, but you're not tiring yourself out with raised arms, either.

Re:Still needs camera interface

[EvanED]

I'm not sure what your point is... A camera recording positions of bright dots is one of the most common motion capture technologies.

It's not exactly like the Wii... the Wii tracks IR emitters, whereas motion capture more commonly uses reflective spheres and a separate light source. Also the camera is stationary, and the light sources are moving in motion capture, whereas in the Wii it's the other way around.

But I would say the two techniques are a lot closer than you seem to think.

drift is an issue

[penguinbroker]

Aircraft have been using this combination of sensors for a while to handle attitude adjustments, however over time the sensors will accumulate minute errors that ultimately compound into larger ones. For this reason, an absolute reckoning system like GPS is always included.

This is a great step forward but does not mean current IR strategies are necessarily old news. The blend of these two systems holds the future.

Re:Still needs camera interface

[eh2o]

IR tracking has zero drift, unlike accel+gyro IMUs, although, after calibration for the local magnetic field, the magnetometer (compass) can provide the necessary correction. One will have to tell it where the TV is, unlike the wiimote which already knows, but it will still work when not pointed at the IR source so there is more possibility for independent 3D motion tracking.

The problem is, those extra sensors are not cheap (currently) -- esp. compared to the stuff in the wiimote. The $99 price projection is likely a pipe-dream or on such narrow margins that Nintendo would never take such a thing seriously.

Note that TFA says this thing tracks absolute position which is an error -- it tracks absolute orientation. It's position sensing will be short-time relative at best.

Re:If Nintendo is smart...

[TheRaven64]

I think the 'next generation' is a bit further off than TFA implies. I recently had the opportunity to play with an experimental device that our HCI department had which worked in exactly the way described in the article. While it was great for things like gestures, accurate position tracking was impossible due to rounding errors in the digitisation of the sensor readings. The reason the Wii uses IR is that the other sensors need recalibration very often, and the IR lets them do this. The Wiimote works better than the device I played with (and costs a tenth of the amount) as long as you wave it in the direction of the IR emitter periodically.

Re:If Nintendo is smart...

[Ichinisan]

The IR function of the Wii is for pointer functionality only. The Wii remote already uses MEMS accelerometers for motion-sensing in the Wii remote and the nunchuk attachment. -Signature