Yes, sorry, using PP.
Had another go today and seem to have made progress, mainly because i have set fusion to 100% GPS**. With just GPS i have achieved a reasonable line following. Still needs improvement and hopefully IMU fusion will assist.
Specs below. They probably look abnormal, perhaps because of the geometry of the vehicle, compared to the normal AoGPS vehicles.
Haven’t got far enough to try UTurn.
100% GPS Fusion
Look Ahead - 3.5m, Look Ahead Gain - 2, Side hill - 0.75, Integral - 1
P - 200, Max - 201 (Unused yet), Low - 20 (Could probably go lower), Min - 35
Counts Per Degree - 15
**
Main Issue : -
In AgDiag, the “From Autosteer” “Heading” was 90 degrees wrong.
I will start another thread for the issue, or search for answers. I had already recently rotated the IMU because of the same issue as above. Now it seems wrong again…
It doesn’t matter if bno085 point nort east west or South, AOG only compare reading with last reading, and use that to adjust heading.
You can’t get a heading when standing still, the GPS must give at least two points with your set fix distance before AOG know a heading.
Update to the above for anyone facing the same issue / interested.
Fusion - 50/50% to 70% GPS/30% IMU. Still experimenting
P200, Max 200, Low 35, Min 35
Look ahead 2.8 to 3, LA Gain 2, Sidehill 0 and Integral 1. Sidehill and Integral to be changed when tested on terrain.
15 CPD
Got a reasonable line following at that. 1m error total, maybe more in certain areas.
Look ahead and low motor could be tweaked more
Steering action is so fast, which in turn moves the GPS and IMU laterally quickly. Hence why low CPD needed. CPD at 55 or 80 (anything >30) for example both produce huge turns and oscillations either side.
Haven’t tried U Turn. Turns with the current setup highlight that the program gets confused by the quick turns/small radius. May need more IMU to support turns
Maybe mounting GPS further back may help to slow lateral movement…but that’s obviously a tradeoff for other factors…
For example with RTK 90% of the time error is +- 1cm with half that time spent sitting on zero.
The worst error i see, usually beside a tree line or coming out of a ditch is 4cm.
I also agree with @Vili 200P is super high. It effectively doubles the error. Try doing all tuning with it at 100P, and adjusting it at the very end.
With RTK your live position should be reliable to 14mm. 1m is a rather large wave.
Though I am with you on the fusion 60 gps / 40 imu is very stable.
What is your steering degrees at full lock? lots of conventional tractors seem to be in that 35-45 degree range.
For testing it is best to use flat concrete if you can that will eliminates imu and mud errors. From my experience if it’s not working on flat ground it will not work in field.
It seams that your motor is too slow to react try higher low and minimum pwm, lower look ahead, try lower integral, do you have rtk fix? Counts per degree seams high, is your was 0 good ?
I helped you setup a base station, having RTK fix is not a given it is constant struggle to maintain it, any hiccups and no RTK fix.
Close to tree or building not enough satellites in view no RTK fix, tablet doesn’t have internet, phone hotspot doesn’t have internet, no cell phone signal in area, bad configuration of AgIO for NTRIP, your home network doesn’t have internet so you base can’t connect, someone borrows base extension cord, rtk2go doesn’t work no RTK fix.
That’s why its good to check if you have fix when problems starts it’s often because of losing RTK fix, AOG will notify for lost connection with caster.
This little unit loves the PP with the antenna out front? (Excuse the old rtk 150 ezsteer, old pic, but antenna location the same for AOG)
Testing Antenna over pivot is a must do for this summer to see any difference.
Just saying to try stanley. It will eliminate some variables. If it is still wrong in stanley, then its probably was set up or motor speed. Stanley, I find useless for uturn or lane shift. PP, in my opinion is the proper set up, and closer to pivot axle, minimizes a lot of the over corrections while figuring out the settings. If you have a set up that works, let’s log it.
I used stenli and it turned out decent, but it was 4-8 cm from the straight line. The GPS antenna is mounted on the roof. Should I move it to the front?
I will list what I think is a problem:
RTK varies. I stopped and watched several times. My position changed by a few cm (between 1 and 6-7 cm), and the IMU showed the same value. When I set up the simpleRTK2B base station board, the antenna was in the yard and now it’s on the roof, but I didn’t adjust the position. I just mounted and turned it on. Works for more than 24 hours. Do you need additional positioning, point averaging or something like that? Can I measure the accuracy of the signal in the tractor, 1 cm or 10 cm?
Attaching photos: https://photos.app.goo.gl/nKa9VjSq4pDqM5nZ8
The compass is mounted in the MMA slot of the PCB, the box is not fixed. With a stationary tractor, when the box moves by half or 1 cm, I get a deviation of 1-4 cm, even if I press it lightly with my finger, I get a deviation of 1-2 cm. I assume you have fastened your boxes with screws?
This point is similar to the second, but there is another thing that worries me and that is that my cabin has a suspension. I made an attempt by slightly shaking the cabin when stopped and got a deviation between 2 and 7 cm.
I just noticed it. When I stop and open Auto steer configuration the window in its lower bar has 3 values "Set: Act Err:", so when I stop the value “Set” changes. I will attach a video to see if it is normal. https://photos.app.goo.gl/berNp4cuLxhevKsN7
The best that appeared on the screen was 4-10 cm from the straight line, as the tractor meandered now 4-10 cm to the left, now 4-10 to the right. This happened with Pure pursuit and Stanley
