2.54 Rocks!

 

Just loaded 2.54 into APM2,  x525 frame, 880kv 10x4.7, sonar.  Not only do the normal modes fly incredibly smoothly, I have now gotten Auto to do what I have been trying for the past year - repeat missions reliably!

I just ran this mission-

3690928764?profile=original

 

many times, all with the same result!  Here are 3 typical flights -

 

3690928813?profile=original

 

It hit the waypoints smack on every time, altitude was great, landed at the same spot always. I usually took control about a foot hi, but I let it land once and it was OK (did not fully stop props).

It flew so smoothly that I finally had a real sense of confidence instead of panic.watching it terrain follow and decide where to go next was a thrill - the robot is alive!

Stable mode is so good it laughs at ground effect - I can fly a half inch hi perfectly stable.  (Param file attached).

The only basic issue left that I am very glad tridge is looking at is loiter - I have played with it a lot, and while I can get a good loiter fairly often it will just zoom way out for no reason and take a while to get back. I bet the gps ground speed or accelerometer feedback might keep that in check.

Once again I cannot thank the team enough - we are seeing some real resolution to a very difficult problem.

 

                                           

jh254-4-21.param

You need to be a member of diydrones to add comments!

Join diydrones

Email me when people reply –

Replies

  • Developer

    great result John!

    btw, it's Jason who is looking really closely at loiter at the moment. He's doing some really good analysis regarding how to use the accelerometers to cope with loiter drift.

    Cheers, Tridge

  • Oh I should also test this, looks realy good

  • 3D Robotics

    Also, the Loiter issues usually have to do with GPS noise; readings can occasionally vary by tens of meters depending on satellite visibility. The way we solve that is with GPS filtering, predictive algorithms and dead-reckoning augments to GPS. It's just math, but is one of the more complex control theory problems out there, so we're experimenting with different techniques. 

  • 3D Robotics

    Glad to hear it! We've still got some tuning to do to get Loiter tuned in better. Some of the dev team are reporting some good results with :

    AHRS_YAW_KP=0.2


    You might also try Loiter_P=0.1 but high values of Loiter_rate_P, such as 3 to 4

This reply was deleted.

Activity