Slashdot Mirror
Using Hacked Wiimotes As Scientific Sensors
garg0yle writes "Scientists are repurposing Wiimotes as scientific sensors to help measure wind speed or evaporation from lakes, among other things. At about $40 per unit, the controller is much cheaper than specialized sensors. The scientists are still considering how to add storage and extend the battery life."
to detect global warming?
Seems like more of an engineering challenge than a scientific one.
“There are probably better ways to measure wind, but it was a day well-spent,” Hut said. “I really felt the need to solder something.”
A day well-spent indeed! There's nothing like spending a day to save a few dollars by not having to buy a specialized sensor. Sounds like my Master's research; why buy good equipment when grad students can spend ages building a poor imitation of it? Still, those days are usually the most fun part of "science" and certainly afford excellent learning opportunities.
So if this is the future...where's my jet pack?
...this equipment can run $500 or more...
The scientific equipment is more expensive because laboratories are willing to pay more, and have the money. Gamers aren't willing to pay $500 for a controller.
Look here: Digikey has 18000 pressure sensors available. I picked one at random, and it can measure pressure up to 115 psi, which is about 60 meters deep in water. It only costs $12. I could make you the serial port/USB interface for like 20 bucks.
Scientists only pay that much because they are willing to pay that much.
I think the pricing of scientific instrumentation is based largely upon the limited number of devices produced. The folks who make sensors really do not care too much about the price and are looking at recovering their development, manufacturing and marketing costs off of very small sales quantities.
A case in point; I work with AMI (SmartGrid) systems for measuring water and electricity consumption. These devices have a surprising level of sophistication, very long battery lives (10-20 years off of a Li-Ion battery) and can store a data-point every fifteen minutes and report it back across a radio network. I "know" the manufacturing costs are down in the $30-60 range for each device. The manufacturers are all anxious to get customers (utilities) to spend their millions on projects to put SmartGrid technologies into cities so the more you buy, the cheaper they get. The data is frequently coming from "absolute encoders" on water meters and less frequently, from pulse encoders that generate a certain number of pulses per 1000's of gallons (the device counts them up, multiplies them by a K factor and gives you a corrected value for gallons of water consumed).
The Nintendo Wi is a good example. How many millions of the Wi devices are made? If they were $250 each there would not be many consumers buying them so they mass-produce and keep the prices low. You see the same effect when you hear about banks of PlayStation 3's being used in clusters for supercomputing.
Tisha Hayes
I'm curious on the non-game advances the Wii, PS3, and XBox 360 has provided for the community.
The Wii advances via it's mass-produced controller, the PS3 advances via mass-produced mini-computer, the XBox 360... um... (need some help here).
...thats the beauty of science, we're not limited to "have to", but more what we could do - "because we can".
In amateur science circles, we also used commercially available TV-tuners as spectrum analyzers, instead of purchasing a commercial test-instrument that cost up to a 100.000 dollars, it could be made to perform pretty close and pretty well with some external circuitry for a few hundred bucks, made it affordable for the radio-amateur, science amateurs, and science students everywhere.
Absolutely LOVE to see people use the resources like the Wiimote like that, excellent!
So yeah - sky's the limit!
What this world is coming to - is for you and me to decide.
The lab I work in uses hacked Wiimotes to study visual pecerption in autism, Alzheimer's, and Parkinson's. They can be programmed with C#, which our PI and another undergrad learned to work with the Wiimotes...the ease of use and the hackability are pretty good selling points besides the price.
There's a few Wii battery packs out there that allow the controller to be powered over USB with a standard A to mini-B cable. Here's one:
http://www.dealextreme.com/details.dx/sku.4978
Of course, if you drain the battery pack faster than you can recharge it, you might have a problem.
Hands in my pocket
it would sell if you had interactive pr0n games...
C|N>K
More specifically, a tuning fork gyroscope.
Apparently it is easy to read with i2c as a standalone part, too.
Yep. That socket in the bottom of a Wiimote is nothing but a proprietary i2c interface connector. You can get accelerometer, button, and joystick input from a nunchuck with a $2.50 Atmel microcontroller. I don't know if they still have them, but sparkfun used to have a little adapter board to let you connect a Wiimote accessory to a .1" pitch 4 pin header.
While a gyro is necessary to actually do full 6-DOF position tracking (otherwise you must assume that you're holding a specific orientation... this can still be good for something like an in-the-air mouse), the Wii system still wouldn't be good for detecting absolute motion without the sensor bar as well.
The problem is that while the sensors are fairly precise as far as measuring the accelerations (if they're anything like the iPhone sensors they're around 0.02g precise), when you try and integrate them twice to get the position, things start to fall apart. Imagine you do a simple up-and-down motion. You get a sinusoidal acceleration curve that when you integrate it once gives you an offset sinusoid to represent your velocity, and a second integration gives a third one to represent your position. However, at the end, your integration to the velocity level comes out to be not quite zero, because those small acceleration errors will mostly cancel out, but not perfectly. This is still a pretty good velocity estimate, since its close to zero. However, as far as your position is concerned, close to zero and actually zero are very different, so you get a constant, growing drift in your position from a small velocity error. The same things apply to gyros, although the math is a little more complex.
Basically if you want to use a sensor as a double integrator it has to be extraordinarily precise, and even then you're going to get some drift that you have to remove every once in a while, or have an absolute position value to keep it in check (kalman filters do a great job of interpreting data from multiple sensors). What the sensor bar and IR sensors do is give you an incomplete but useful reference on position and orientation that you can use to keep that drift in check. Adding the gyros definitely helps a lot too, but you still need the sensor bar to keep drift in check.
In my experience, doing some sensor systems with cheap sensors and expensive sensors, the difference is that cheap ones can be fast or accurate, while good ones can be both. ... Its really a matter of knowing what you need. In many cases a cheap sensor works really well.
Using cheap sensors can allow you to validate your hypothesis. This can in turn help you justify the additional expense of calibrated and more precise sensors to accurately describe the relationship observed. (You can also use spares of the cheap sensor equipment to play games and blow off steam while the experiment is running; Mario Kart anyone? :)