Using only GPS hold (no altitude hold) my quad decided it needed to be somewhere else in a hurry.
It travelled 15 metres forward and took some quick reactions to avoid a crash. This has happened before and the last time it did this the result was a crash - no damage thankfully.
Has anyone else seen this kind of behaviour?
Thanks Rich
Replies
First post here :) Allow me to give my qualifications. I was a WSN-5 tech in the US Navy. Since then, I've studied inertial navigations as a hobby.
There is one big, bold problem with using GPS strictly as a means of holding the copter in place. If your GPS gets a "better" fix, it can suddenly snap to a completely different location. This would cause your copter to race off like that. This could happen because you have trees or buildings blocking data from certain satellites while the copter is at a low altitude, then the GPS resolves those satellites when it's acended, causing a different fix to be reported.
The solution is to filter your GPS data and your IMU data together. The IMU provides acceleration data. The GPS provides position data. Therefore, if your IMU reports no acceleration, and your GPS has shown through a history of fixes that it's in one position with no velocity, then your copter isn't moving. A sudden snap in position will not be corroborated by IMU data. At worst, your copter should slowly move to the new position as the filtered position slews based upon repeated GPS data.
Yes, I am also having a hard time tuning the Position Hold PID values...my quad is a bit heavy and I had to adjust the stable PID's quite a bit to make it stable..now my problem is Position Hold...the best I can get is the quad flying large circles around the set point.
More tuning is what I probably need.