Correcting accelerometer mounting angles

[Hemib]

Hi all!

I'm working on a system that uses an accelerometer to measure pitch. I know how I can calculate the pitch angle from the accelerometer readings, but I've another problem.

At the start I never know how the accelerometer is mounted. So normally on at the start the accelerometer should give me [Ax Ay Ax] = [0 0 -g]. But this is not the case, because my accelerometer is mounted randomly at the start. So it returns instead [Ax Ay Az] = [a b c]. How can i know calculate the right angles of orientation when moving?

I've also have access to gyroscope readings (also randomly mounted, but always aligned with the accelerometer). Maybe a solution is to find the orientation angles first with the gyro (I think that for a gyro the mounting is not important, correct me if I'm wrong!) and then correct the accelerometer.

So I need something like this:
[ $$\theta$$ $$\Psi$$ $$\Phi$$] = T [a b c]

- theta, psi and phi are the real orientation angles
- T is the transformation matrix
- a b and c are the wrong accelerometer readings

Thx!

 

[mender]

Rotate the accelerometer 180 degrees and take a second reading. The errors will cancel.

 

[Hemib]

The accelerometer is fixed mounted, but in a random position. It's not possible to turn is 180°

There has to be some way to calibrate this.. Maybe with a reference to a GPS signal?

 

[russ_watters]

Why can't you use the accelerometer output to calculate the initial angle, then subtract it from all later calculations?

 

[Hemib]

How do you do this? Problem is that it's possible that in the initial position the angle I want to calculate is not 0°.

 

[mender]

What's the application? If it's aircraft, most aprons are flat enough to get an initial calibration.

 

[Hemib]

It's for calculating pitch angle of a vehicle

 

[russ_watters]

Well you'll need something to get you a calibration, either starting on a level surface or measuring the initial angle by another means and comparing it to the angle the accelerometers measure.

 

[Hemib]

Hmm.. Ye, was thinking about using GPS for this.

Btw, is a gyro also affected by his mounting position?

Thx for replies!

 

[ACPower]

I think you can just record the initial (level) value for the 'down' vector, then just use the dot product of that vector with later down vectors to get the cosine of the angle of the vehicle. For a car you can assume that all of the angle is front-to-back tilt.

 

[Hemib]

Ye, then you have the angle with respect to the initial position, right? But what if the car starts on a steep hill?

 

[Darryl]

Im sorry, but I am a little confused in what you are trying to do, you are using an accelerometer to measure pitch, by referencing it to gravity, so if the vehicle is not moving the accelerometer will read either zero (if horizontal to g) or 1g if verticle to gravity.

But if the car is accellerating and is pitched, there is (I think) no way to differentiate between the two forces.

In other words the sensor will not know if you are accellerating or changing your pitch.

The only time, I can see that system giving an accurate reading of pitch is when the vehicle is not accellerating either standing still or traveling at a constant speed.

What information is provided by your gyro, if it provides precession information or some other way of seperating (and then subtracting) the acceleration and pitch data.

Im also (only guessing) that such a system would intruduce measuring errors I would expect a gravity based system, with both pitch and accelleration to give different results depending on if you are going up hill or downhill.

Im thing something like mounting the accellerometer on a gymbal, but I stall have trouble see how the system would be able to differentiate between accelleration and tilt, without fixing one of the values seperatly..

 

[Hemib]

Its indeed impossible to detect if the change of pitch is due to vehicle acceleration or gravity. But that's not the problem. The problem is that accelerometer axes or not aligned with the 'measured' axes. So how can I get the right pitch angle from a wrong mounted accelerometer.

 

[mender]

What other vehicle sensors can you tie into? If you can access the vehicle speed sensor (VSS) you can do a direct calculation of the acceleration rate and calibrate from that. Maybe.

 

[Hemib]

Ye, I have access to those sensors. Thx

But that doesn't solve the mounting problem

 

[Hemib]

Im also a bit confused with the gyro. Is a MEMS gyro also dependent of the mounting position? because otherwise the gyro would give me the right angels, so i can calibrate the accelerometer

any idea what sensors would do the job? would this one be good? www.sparkfun.com/commerce/product_info.php?products_id=9268