• Registered: 2008-12-26
  • Posts: 1

Topic: Possibility of getting Chumby's position using accelerometer

Hey guys!

First I want to thank you for the good work on the Chumby, I really love it!

Now to my question: A friend and I discussed about getting Chumby's position when you move it around using the accelerometer. I've already read a little about the accelerometer itself and the driver who delivers the data. But is it possible to accurately calculate the position with these values? If yes, which values are the correct ones to integrate? I don't really get what all these values stand for...

Thanks in advance for your reply

Danny
Computer Science Student
University of Basel, Switzerland

  • Registered: 2009-01-07
  • Posts: 2

Re: Possibility of getting Chumby's position using accelerometer

For positioning, I think it would be pretty hard.  Assuming you could sample frequently enough not to miss any big movements, you would still have to compensate for gravity showing in the readings.  If you require the Chumby to always be correctly oriented, you could cancel it out easy enough, but otherwise it probably takes some fancy math to keep the reading for gravity aligned even when the chumby isn't.  Hope that makes sense, I had this idea once too, but didn't really want to delve into the vector/angle maths. =P

  • Registered: 2007-07-04
  • Posts: 64

Re: Possibility of getting Chumby's position using accelerometer

You would need to derive position and velocity from the acceleration.  Make a thing that takes the chumby and once you hit a zero button it says this way is up.  We did this on our robots one year and drift was an issue and if the chumby changes position when sampeling it will mess up your results.