Auto heading

[Marco Trapanese]

Hello,

I need to develop an auto heading feature for a small underwater vehicle
(ROV).
On board there are a rate-gyro and a flux-gate compass.
I was able to use the rate-gyro for short-term heading stabilization,
that is it tries to maintain the current heading in presence of external
or internal disturbances.

Now I need to add the long-term heading stabilization, in other words an
auto heading capability. The user sets the desired heading and the
vehicle should reach and maintain that.

I tried with a standard PID using the compass signal as feedback but
it's too slow and noisy for a fast recovery of the bow direction. After
some tests I ended up with a trade off: with a weak control loop I have
a smooth but inaccurate behavior, with a strong one a more accurate but
instable behavior.

I'm sure I could improve the whole thing if I use the rate-gyro in
conjunction with the compass. But I don't know how to mix both data.

In summary:

* the set point is a magnetic direction
* the inputs of the system are the current magnetic direction and the
current rate of turn
* the output is a value proportional to the rate of turn needed to reach
the set point.

May you suggest me a way to follow?

Thanks
Marco

 

[[email protected]]

http://tom.pycke.be/mav/69/accelerometer-to-attitude

You'll have to do quite a bit of reading of his blog to find out what
you need to do.

Steve

 

[Rich Webb]

Hello,

I need to develop an auto heading feature for a small underwater vehicle
(ROV).
On board there are a rate-gyro and a flux-gate compass.
I was able to use the rate-gyro for short-term heading stabilization,
that is it tries to maintain the current heading in presence of external
or internal disturbances.

Now I need to add the long-term heading stabilization, in other words an
auto heading capability. The user sets the desired heading and the
vehicle should reach and maintain that.

I tried with a standard PID using the compass signal as feedback but
it's too slow and noisy for a fast recovery of the bow direction. After
some tests I ended up with a trade off: with a weak control loop I have
a smooth but inaccurate behavior, with a strong one a more accurate but
instable behavior.

I'm sure I could improve the whole thing if I use the rate-gyro in
conjunction with the compass. But I don't know how to mix both data.

In summary:

* the set point is a magnetic direction
* the inputs of the system are the current magnetic direction and the
current rate of turn
* the output is a value proportional to the rate of turn needed to reach
the set point.

May you suggest me a way to follow?

Google "Kalman filter" to get you started.