Accelerometer

Hoffa

Hi

I´m using a basic accelerometer (memsic 2125) to calculate the distance that the robot have moved. I have a proplem with that the values keep changing now and then even if the accelerometer stands absolutely still.

I hade an idea that the robot could callibrate the accelerometer when it stands still. Put to know if the robot is absolutely still i need to use the eccelerometer...

Have someone had the same proplem? Do someone know how I can fix this? Are there an another good way to say if the robot isn't moving (It's a quite big robot so it doesn't work to look just if the engines are moving because of the skid risk)?

Please help me...

/Hoffa

Bergamot

Hoffa said...
I´m using a basic accelerometer (memsic 2125) to calculate the distance that the robot have moved. I have a proplem with that the values keep changing now and then even if the accelerometer stands absolutely still.

Since acceleration is the second derivative of position (with respect to time), you're going to get some serious quantization error from integrating the data twice. That's not what's causing your current problem, it's just something to watch out for.

EDIT: Which values keep changing, the acceleration, or your calculated position? If it's the latter, then quantization error could very well be the problem.

Paul Baker

Any system of measure is going to have noise in the system, there are many causes for this noise but it is always present. The level of noise as compared to the real signal is called the Signal to Noise Ratio (SNR) if that quantity is large (ie the noise is relatively small), you can truncate the measurement to ensure that all significant figures belong to the signal component thereby eliminating the portion of the measurement affected by noise. If your SNR is low (noise is a significant factor), then you need to employ much more complicated techniques such as Kalman Filtering.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Paul Baker
Propeller Applications Engineer

Parallax, Inc.

Post Edited (Paul Baker (Parallax)) : 3/14/2007 5:39:19 PM GMT

Hoffa

I thought of a system that calibrate the accelerometer everytime the robot stands still. Does someone know a good way to check this (without using the accelerometer)?

simonl

Would you be able to have the robot stand still for (say) the first few seconds after startup? If so, you could average the accelerometer's readings over that period and use the result?

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Cheers,

Simon

BTW: I type as I'm thinking, so please don't take any offense at my writing style

---

www.norfolkhelicopterclub.co.uk
You'll always have as many take-offs as landings, the trick is to be sure you can take-off again ;-)