Replies to This Discussion

ModeratorPermalink Reply by Chris Anderson on November 18, 2009 at 5:34am

That is correct. You also would not be able to correct for centripetal acceleration without a velocity vector

• ▶ Reply

Permalink Reply by Daniel Wee on November 18, 2009 at 6:33am

Thanks Chris. This brings me to the RollYawDemo in the gentlenav directory - that seems to be doing with only 5DOF. Does that mean that if I am purely interested in roll and pitch estimation, it can still be done using the DCM algorithm. This is where my confusion is coming from - the math looks like it requires 3-axis of correction to solve for all the equations, and so in theory a 5DOF IMU without GPS input will not allow orthogonality to be enforece - and yet there is that RollYawDemo. Is that demo a proper implementation of the DCM?

In other words, if I take that 5DOF RollYawDemo up in flight, will it still work properly.

Thanks,
Daniel

• ▶ Reply

DeveloperPermalink Reply by Doug Weibel on November 18, 2009 at 7:50pm

Daniel

Maintaining orthogonality is not the problem. Normalization and enforcing orthogonality is simply math in the algorithm. 3DOF gyros are required for all but shallow bank angles to maintain a converged solution without overly relying on correction. 3DOF accelerometers are required for gyro drift correction, as is GPS

• ▶ Reply