The Wii's MEMS Inventor on Future Technology

eldavojohn writes "IEEE Spectrum is running an article on the inventor of the motion sensor that the Wii uses. The microelectromechanical system (MEMS) gives Wii its core ability to sense motion in the controller. What's really interesting is where Benedetto Vigna wants to take this technology. He has plans to make the sensor smaller and tougher, and hope to place it inside of things like shoes, textiles, and medical devices to aid in data collection. He continues, 'Then I want to make a three-dimensional gyroscope, to measure rotation around three different axes. Today, such products are quite big, a cube 10 centimeters on a side. We want to do this in less than a 30-millimeter cube, to serve as an image stabilizer in cameras and to track a person's position in the intervals when he can't get a GPS signal.'"

This is cool, very cool...

[zappepcs]

I have not played with a Wii yet, but knowing something about robotics I can say that if they manage to get a 3D sensor set working, and cheaply, it will advance a gazillion projects. Knowing how and when to place mechanical effectors and movement of devices is a terribly difficult problem generally. This type of sensor will help do that very effectively.

This can be used in conjunction with other sensor systems to do things like create a lawnmower robot that doesn't just wonder around till you turn it off. Being able to manage calculation of 3D space is very intensive, but doing so lets us get one step closer to the robot maid and other cartoon dreams of days gone by.

Its not just for games. Most of the semi-successful DARPA grand challenge vehicles used a similar device for navigation support. The reality of a car that drives you (in Soviet Russia) to work without any intervention from you is getting very close. Inertial navigation (AFAIK) relies on 3D motion tracking to determine the motion in between points of absolute (or relatively absolute) positioning data. So, in between GPS readings, inertial navigation estimates where the robot/car/vehicle is in relation to previous GPS readings. I've seen robots do this already, its just not cheap enough for everyone. A small R/C sized robot can travel 1/2 mile and return to its starting point with high accuracy despite obstacles using inertial navigation. This can be applied to a lot of systems.

Re:Position tracking?

[aweinert]

GPS for gross location, MEM gyroscope for tracking small changes in velocity, position, orientation, etc., and when GPS fails.

already done

[jimmydevice]

The Crista Inertial Measurement Unit is a very small three axis inertial sensor that provides high resolution digital rate and acceleration data via serial interfaces. It uses MEMS gyroscopic rate sensors and accelerometers mounted on orthogonal axes to provide 300 /sec rate and 10G acceleration data. Small (2" x 1.5" x 1", 37g ) http://www.cloudcaptech.com/crista_imu.htm [cloudcaptech.com]

Already there

[truckaxle]

FTA "Then I want to make a three-dimensional gyroscope, to measure rotation around three different axes. Today, such products are quite big, a cube 10 centimeters on a side."

There are such devices now that are compact and capable, such as...

http://www.microstrain.com/3dm-gx1_specs.aspx [microstrain.com]

I worked with this device last summer implementing a vehicle flight path recorder. It not only has 3 rate gyro's but three 5 mG accelerometers, a compass and processor that implements navigational processing and outputs earth-frame quantities via a serial connection.

Size: 42 x 40 x 15 mm

Some links to the datasheets

[Anonymous Coward]

Links to the Analog Devices pages:

Accelerometers:
http://www.analog.com/en/subCat/0,2879,764%255F800 %255F0%255F%255F0%255F,00.html [analog.com]
(Mostly 1 or 2 axis; the only 3-axis one is the one used in the Wii. It costs $5.45.)

Gyroscopes:
http://www.analog.com/en/subCat/0,2879,764%255F801 %255F0%255F%255F0%255F,00.html [analog.com]
(All available parts are 1 axis. Costs from $30.)

Here's the fun stuff. This not-yet-available part:
http://www.analog.com/en/prod/0%2C2877%2CADIS16350 %2C00.html [analog.com]
combines a 3-axis gyro with a 3-axis accelerometer, and is close to what the author is referring to; it's a cube about 23mm on each side. It looks like a great product, if the price is right.

Re:No Thanks

[LiquidCoooled]

The most frustrating example of this is the final putt on the golf course, the one which is about a metre away and you need minimum power.
You find yourself doing some random body shake to try to get it registering.

Half the time this ends up with the ball fucking off half a mile away and landing in the pond.

Re:it's already here

[Anonymous Coward]

They aren't up on par with inertial-navigation-grade systems using fiber-optic gyros, so their drift tends to make them unusable for long-term navigation. You wouldn't fly a cruise missile or an ICBM on them.

I worked on the inertial sensor designs for the US's strategic grade guidance systems at Draper Laboratory. We only use MEMS, no fiber optic stuff. Same technology that is in the Wii except much more accurate.

Re:No Thanks

[crazed gremlin]

you need to hold it so the left side of the wiimote faces the sensor bar...I had the same problem until my friend enlightened me.

10 centimeters!

[arsenix]

This guy needs to spend 5 minutes googling for IMUs (intertial measurement units).

For instance, this unit:
http://www.memsense.com/products/mag3.asp [memsense.com]

There are a million of these out there...

Has three axes of accelerometers, three axes of rate gyros and a three axis magnetometer... all in a package that is .7"x.7"x.4".

Re:Clarifiation.

[smallfries]

The principle is the same although the practice is different. Each integration accumulates error, so adding the extra layer degrades the performance.

These ideas aren't new and have been knocking around for a while. The article sounds a little like hype / ego-wanking, but then again IEEE Spectrum articles normally are. There is a ton of work on "sensor fusion". The basic idea is to take several low-grade position sources and then fuse them together to create a (hopefully) high-accuracy position source. The robotics and wearables communities have been looking at this for many years. One nice approach is combinng the sensor inputs in a Kalman filter which does actually create a higher accuracy signal than any individual source.

As far as the claims about 3d gyroscopes being the next big thing when they are reduced in size - we saw a demo of a commerically available product about two years ago. It is a 1cm cube that intergrates several accelerometers and gyroscopes to provide a dead-reckoning position source that is accurate to within 5cm. It was very impressive, although the cube cost several thousand pounds. It would be pretty amazing to see Nintendo pick up on something like that.

Re:This is cool, very cool...

[xenocide2]

You can buy the same part the wiimote uses for about 8 dollars from digikey. Or a better one. I don't know where you get your numbers from. It's not as cheap as Nintendo gets it, but if you're considering a serious deployment you should just ask for engineering samples. Intertial navigation just doesn't work. Sensors aren't perfect, so you lose some position that way, and you can only sample so fast, so any short variations can potentially be lost, or brief peaks extrapolated for much further than they really were. And even then, the ADC that samples the voltage isn't perfect, so more accuracy is lost there! Nintendo's own engineering team states this has been their experience, in an interview, which is why they have the sensor bar. It is not, as the article implies, only to gather the initial position of the tv relative to the wiimote. It's there because inertial navigation deviates over time. You can easily correct the deviations with say, a GPS device, or the sensor bar, though.

I don't understand why you're crediting the Wii with making the parts so cheap - it was STmicro that put together the factory, and STmicro was selling them cheaply before the Wii was huge.

Motion sensors not unique to Wii

[aero6dof]

The motion sensors aren't the unique part of the Wii. Sony's controller has the equivalent motion sensors. The unique part of the Wii is the combination of the motion sensors with the IR bar tracking to give you a non-drifting reference.

By themselves, the motion sensors will get further and further off position. For example, if one turned right 90 degrees and then returned, the motion sensors by themselves would cause you to calculate a position not-quite matched up your original - and the more you move the more the reference will drift as measurement errors accumulate. With the IR bar, the reference can be corrected so the controller can stay oriented correctly vs the screen.

This is why Sony's controller is a very poor substitute for the Wii controller.

Re:No Thanks

[denebola]

Personally, I find the easiest way to make the short putts is to reach a threshold speed but increase the angular movement. Not like a real put which has fairly small angle. I use ~40 degrees with a slowish acceleration.

HTH.

Biophysics: Antennae as Gyroscopes

[dimeglio]

Then I want to make a three-dimensional gyroscope

Benedetto Vigna should read this report http://www.sciencemag.org/cgi/content/abstract/315 /5813/863 [sciencemag.org] about how moth are able to manuver so well in space. Their antennae are a small, very small device which does the job amazingly well. If first heard about this on Quirks and Quarks http://www.cbc.ca/quirks/media/2006-2007/mp3/qq-20 07-02-17d.mp3 [www.cbc.ca], a science radio programme.

To fly we observed how birds did it, then instead, built wings as used in airplanes today, instead of wings like birds have.

Now if only I could get my hands on a Wii...

Re:Snake oil technology warning

[Speare]

Erm, except to do it your way, each of the microexposures have to be quantified before they could be shifted and averaged, and when you quantify a high-gain low-photon data set, you INCREASE the amount of noise in the final data. Also, since it would take some time to decide how far to shift each of the microexposures, you're now taking longer to get a total of 1/30sec of actual photon-catching exposure, and fast-moving objects would appear to be stuttering along instead of smoothly blurred.