Thank you. I will go out shortly and try these to see how they work, then see if I can fine tune from there and I will get back to you with an update later this afternoon or tomorrow morning.
Hope you make some good progress. It’s pretty clear why you’re having issues, but I think they are all solvable.
I only got a few minutes of testing in before the rain started back up, but here is what I have noticed so far and done so far.
I changed all of the above parameters like you had suggested. I disabled the internal compass like you had suggested and I still noticed that on the mission planner screen the rover symbal was spinning in circles. So I disabled compass #3 and just had it use compass #2 which seemed to fix the “spinning” problem. But now the direction sometimes points the wrong way on the mission planner screen, which I assume can be fixed by a new compass calibration?
Then I noticed that it was pulsing pretty badly along the straights so I changed the speed P parameter from 0.35 to 0.2 and I left the speed I parameter at 0.2.
Because I changed the parameters that I just mentioned above, I had stopped the mission, then I went to continue again and the rover would not drive towards the waypoint. I suppose this is because of the compass heading issue mentioned above. Would you agree with this?
That seems like a reasonable assessment.
Next time you’re able, try the COMPASS_LEARN procedure (automatic offset calibration) while driving in those same large figure 8s I mentioned before.
Advanced Compass Setup — Rover documentation (ardupilot.org)
That should help set a baseline. Once it completes, reboot to ensure that a fresh log is created, and drive in figure 8s again for a few minutes. We’ll use that for MAGFIT, which should give you the best possible chance of success for a good yaw/heading source.
Perfect, with weather permitting, I will try again in the morning. And thank you again for all of your help thus far.
Just out of curiosity, that bigger rover that I had mentioned before (sand colored) does not have any compass enabled, along with a single F9P GPS in the front/centered of the frame. But I was able to achieve accurate results, along with tracking very straight lines (although it did take me A LOT of tinkering to get it right). Does this seem possible? I have run it 3 or 4 times so far with very good results.
I’d like to see a log to confirm what you’re observing. It’s possible that the high quality GPS is providing enough course updates to keep the GSF fallback from drifting too quickly. Having no empirical heading/yaw orientation source is almost always a bad idea, especially when you’re trying to achieve precise results.
I will also try to get a log tomorrow on the larger platform as well. Do you think enabling a compass on this platform would throw off the tuning that is on it now? Or simply improve navigation overall with no overall change to the tune?
Share a log. A compass or moving baseline yaw source is almost always preferable.
I will try to get a log from both machines over to you tomorrow morning. Thank you again!
Update: I do not have a log for it because I got carried away and ran the robot for a few hours this morning since it was working well (the larger platform- sand colored) as I switched from EKF2 to EKF3 and it made a huge difference along with enabling an external compass. I will plan to test the smaller black unit this afternoon and attach a log and update of my progress.
@Yuri_Rage I apologize I have not gotten back to you sooner. We had a mechanical malfunction that actually caused the gears in the gearbox to break and rendered the rover useless. I have just gotten a new gearbox installed earlier this week which allowed me to continue testing. I switched back to a previous version of mission planner as I did not know how to slow down the oscillations that were being caused during auto missions. I will post a new topic on the correct page and hope for your continued support there. Thank you for all of your help.
1 Like