Edit
Yes, I ran it without the antenna and it was the same.
Antenna virtual position should it not be 0âŚclutching at straws though as I havenât got a working setup yet (everythingâs gone wrong this year).
No, virtual position is the virtual center of the vehicle. Theoretically, if the antennas were 100 cm apart and centered with the wheels, the virtual would be 50 cm right so to act as a single antenna in the middle of the tractor. Set AOG to 0 and adjust virtual in esp to center the tractor in same tracks coming and going.
Edit
I thought that the north south being off would just be an imperfection to live with, but where I sowed turn rows back and forth in a âUâ shape due to starting along a fence in some no till, proved to be tough.
1 Like
I would try setting the antenna virtual right position to -63 and roll to OGI. Then try adjusting the virtual right position small amounts say -62,-61,-60 and see what the effect is.
I thought the physical right antenna was position so you have to move it to the left (negative). I have mine set at -60 and it seems to be working well now but I have had similar problems. I found there is something strange about the antenna virtual right position, on my set up, it is very sensitive and moving it small amounts caused large overlaps. I set the virtual forward position to 0, I have not experimented with this and Iâm not sure why it is needed.
I agree. I thought it was the right antenna as well, but i lost RTK signal the other day and the tractor moved to the right and centered the left antenna over the line. When it came back, it jumped back to the left and centered the cab again. Confusing, but darrenjlobb said earlier that he had some problems with offset as well. I will try the negative and report back.
P.S.
I have my esp stack with Rover on top, Moving Base under Rover, with right antenna to Moving Base. Is this correct ?
Edit
The virtual forward is just to get the antenna closer to the steering tires, for a smoother and more accurate positioning of the vehicle. It works well
Edit 2
I tested with negative numbers, and it leaves a gap between passes north and south, and offsets east and west the same as NS was before
It just occurred to me. Are you certain that your right antenna is connected to moving base and not rover. If they are backwards that may be the reason for the negative numbers in offset.
Just a thought
My left antenna (driverâs left hand side) is for heading providing the Relposned message. My right antenna is position providing the NAV PVT message. My upper board is the left heading board and my lower board is the right position board. I made the connections using the Pixhawk connector as described in this thread on the 15 May.
I am the same except stack. Please letâs try to figure out what is different ? I have aog vehicle settings 0 for antenna height and 0 left right offset. Only wheel base and distance in front of axle have values. Same for implement, no side offsets. I did switch back to imperial units, as it doesnât seem to be affecting my offset problem. At startup, I get heading pretty quick without any vehicle movement, and roll works, but is a little lagging. I can speed it up un the code a little later. Left roll is negative as well. I donât know of any other setting that might help ?
The settings you mentioned are the same on mine. Looking at your screen shots, in messages to send to AOGs you have everything selected, I only have PAOGI selected, and the factor I have left as 1.20. I am using NTRIP and WIFI but I donât think that should matter. Running out of ideas, sorry. I would keep trying different numbers + and - for the right position even if they donât correspond to the physical setup and see if you can reduce the overlap.
Thank you. I did select everything just to monitor when I was setting it up. I think I can get NS and EW close, but I have concerns about being able to plant. Will post results.
Edit
I had no success with virtual right, so I started changing virtual forward, just grasping. With -215 cm, the east west, and north south track good, but the 45 degree between them were all over the place. Naturally with the antenna that far behind. Then I closed the web interface and everything went back to good EW and off track NS. I restarted the system and web interface shows -215 still but still off track.
Have calculated earth radius fo my lat, and changed it in the code. Plan to upload and try, but I fail to see how it will change calculations much.
Just to see, I drive the entire loop in reverse after I drove in forward, to keep the antennas in the same order, and it tracks exactly as it did in forward.
Yellow is clockwise, blue is counter. When I drive counter, the Nw and Se corners are close to actual. Clockwise, Ne and SW are correct.
Hi,
a possibility to check the directions the virtual antenna moves would be:
- stay at one place and set all to 0
- set a flag in AOG
- put in 1000 cm = 10 m in webinterface
â so you can see in wich direction AOG moves
Thats the way I tested it while coding.
I changed a sign long time ago, hoped to solve the E/W Greenwich problem, but I didnât get a report if it helped, so maybe the roll moves in the wrong direction.
An option to test this might be to put in values for roll correction when standing still an see AOG moving.
Thanks, I will try it. If everything is correct the vehicle should move opposite the direction of the offset, correct ?
What is your suggestion for the value in aog for antenna position in front of rear axle, actual, or projected from esp ?
I downloaded your latest from github from a few days ago to use gga. The lat lon math looked the same. Was thinking about trying it though ?
Edit
Matthias, have you looked at this post Profile - StefanA - AgOpenGPS
This honestly looks like EXACTLY what mine was doing the other day, and was purely the incorrect +/- for antenna offset in AOG. Although if your using matts code i guess you use offset in ESP and leave AOG at 0 offset? I am sure my AOG offset changed from - to + by itself as worked perfectly for a long time and randomly broke, took me a while to figure it out!
Okay, do you set the virtual forward to position it over the front tires. Otherwise it would always be hunting for the line, since there is no value in aog to trig the steering amount ???
I changed the earth radius to my position.
Very definite pattern to the radius value
Yellow and green are with original download, and offset as close as I could get. East and west return to same set of tracks, north and south are still half pass off.
Orange and blue are with my radius value at elevation 63709710. Orange is clockwise. East and west are a full path outside, and north south full pass inside. The opposite applies for blue going counter. As Darren mentioned, these were ran with same numbers in esp and aog all zero except wheelbase.
With your original radius, I could get within about 1/8th of the swath both ways with a high positive number in virtual right.
Edit
I did try what you said with the flag and setting still.
Positive value in virtual right moves vehicle left of flag. Positive in forward moves vehicle back away from flag.
Edit
Radius value is incorrect. Disregard please
Perhaps try setting virtual forward position to zero and then try the right position with + and - values. Others have said, if you are using the Stanley method together with good heading readings then the antenna position doesnât need to be close to the front wheels. I have my antennas over the rear wheels with zero forward position in the web interface and it works well. I believe it is more important for Pure Pursuit to have the antenna over the front wheels.
I donât use any virtual forward / back⌠The antennas are mounted at the front of the cab, and that is the distance from axle I use.
I use Franz code not Mattâs but should be the same.
Thanks Rich, I have set virtual forward to zero, then set virtual right. I can get closer to correct north and south, but then east and west come off center also. The green and yellow in the picture above are where I did that. I am using pure p instead of Stanley. I havenât been able to get Stanley to drive well. It seemed to lag. Probably my lack of understanding how to adjust it. I am running hydraulic valve, if that matters.
Thanks for that confirmation Darren. I had vehicle forward in aog at zero, and it would not steer well at all. I set it back to actual, and very big improvement. I set virtual to zero verses over the front axle, and it seemed to be closer to the line north and south, but it didnât drive as accurate.
Edit
As far as my testing with the earth radius in the code goes, I changes the numbers to my actual lat with elevation. It did track a lot closer north and south, but had some odd things in aog when I set up curved tram lines. Also had concernes about accuracy with the lat radius being even greater distance from polar radius. So I loaded back to Mattâs code and started over. Have it back to driving good east and west. Have the virtual over the front tires, and aog forward to actual distance(45 inches), wheelbase in aog, and virtual right at 63cm, which is where it repeats back and forth, and also have actual antenna height set in esp. As long as I plant east and west I think it will work, but pretty discouraged as I have tried every offset + and - with no usable improvement.
I find stanley works far better for me all round, if its not happy, are you sure your corrections are coming in fast enough, what hz rate does AOG display the GPS / RTK fix at? anything under 8hz will be hopeless i find, and never stop occilating, fast update rate is extremely importantâŚ