accelerometer

WaldoDTD

Hey,
I am interested in building a robot using the Mesmic Tri Axis accelerometer to figure out the orientation of the robot relative to a base point. Would an accelerometer be the best tool to use for this application or should i look into something more along the lines of a gyroscope? thanks!-Waldo

Bruce Bates

Waldo -

Neither of the devices or sensors you mention has any inherent way of knowing where it is relative to a distant point (the "base"). You may wish to investigate one of the methods shown below if you want to know anything more than that the base exists nearby:

1.·Exact Tracking Methods

··· A. Accurate wheel encoders and a third wheel with·locomotive command tracking

·· B. An accurate compass and an accurate time base


2. Localization·Using Communications Methods

·· A. Radio frequency means·(with or without triangulation)

·· B. Infrared or visible light·means (with or without triangulation)

·· C. Ultrasonic or audible sound means (with or without triangulation)

·· D. Global positioning means (GPS)


I'm sure there are other means but those are the ones I've used or investigated.

Regards,

Bruce Bates

WaldoDTD

What if given certain initial conditions i just took the integral of the function used to find the acceleration two times to get position?

Bruce Bates

Waldo -

At this juncture it might be wise to review what it is that an accelerometer actually measures, and how they are used. Here is a simple explanation:
http://en.wikipedia.org/wiki/Accelerometer

Although accelerometers can be used as tilt sensors, other than that, and because they rely on sensing gravity, there is really no real directional vector that they measure. Thus, they would be fairly useless to your stated purpose "to figure out the orientation of the robot relative to a base point".

Regards,

Bruce Bates

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔

Paul Baker

In theory position information can be achieved via accelerometer data, however in practice it is a very difficult because accelerometers have a fair amount of noise. You may be able to do some predictive analysis using a Kalman filter to achieve a better result, but a simpler means will likely produce better results. Use wheel encoders to measure the distance traveled and a compass to get direction traveled. By using trig functions you can determine the distance and direction back to the starting point.

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

Parallax, Inc.

JWood

Hello Waldo,
Accelerometers have a fair amount of noise and gyros have a fair amount of drift but together they make magic. Using a kalman filter you can combine these measurements to track the angle relative to the gravity of earth.

You can find a spin object at http://forums.parallax.com/showthread.php?p=696561 that take’s in the measurements from an accelerometer and gyro via a 12 bit ADC. It then uses a kalman filter to figure out the angle of the mounted sensors.
·