Hoeing organic crops - climbing up the AgOpenGPS learning curve

Build Reports Project ideas - in progress
1

Time to draw a coherent picture.
My current process looks quite simple:

  • Seeding standard crops (wheat, barley, oats, spelt, rye, beans…) with an old 6 m trailed Horsch DSS at 21 cm row spacing
  • maybe pregerm harrow with tractor in the trams
  • Following with the interrow hoe some weeks or months later
    Hoe shears are 14 … 16 cm, so we have 2 … 3 cm survival margin for the crop - at least in early stage.

The challenge is complicated by undulating sidehills of up to 20 .. 30 percent and most field boundaries far from straight lines.

For hoeing, I still use a venerable Fendt F275GT with the hoe behind the front axle, in direct view.

e0b1e55faa9b5a49a15933269721707baa4eb2ef_2_690x387
e0b1e55faa9b5a49a15933269721707baa4eb2ef_2_690x387690×387 89.9 KB

Precision steering I considered “nice to have” the last years. But sadly , the last F275GT was built in 1984 - more than 40 years ago. There’s no serious successor for it. Some Swiss and Italian manufacurers offer hydraulic driven GT for vegetable gardens in level lands, but their own sales force warned me never to expect the performance, endurance and reliability of a mechanically geared Fendt in continued large hilly operation.

Similiar with camera control: At least 15 years ago, my stated performance goal of 8 km/h at 20 % side hill with 40 cm crop, distorted by wind, was answerd as “ridiculous - impossible” by the salesmen.
At my response “no problem with a GT” I got “Oh, when you have a GT, then there’s no way to sell you a camera…”

In the first step, for seeding only, I’d hope to solve the problem of row spacing between adjacent tracks by reliable RTK cm-level-precision.
When e.g. I target for a 25 cm spacing, I might run a 15 cm knive that does not conflict with the next row at 20 cm and still covers 30 cm spacing at running back and forth.

Here are pictures and discussions what happens when I exceed this margin.
“Account for reduced effective tool width at side hill - #9 by wjr”

To summarize and complete my expectations:

  • ± 5 cm precison in 99%. If we take that figurea as 3 sigma, we head for a sigma of 1,5 cm
  • reliable in side hills and up to 3m from tree’s canopy
  • correction of sidehill and curve errors on adjacent tracks, if they exceed expected errors of 1.5 cm target sigma
  • persistence and repeatability of the tracks for subsequent operations
  • interactive desktop preplanning of the tracks to balance wedges at nearly-parallel headland, track distance errors and curvy tracks due to undulation and curvy field boundaries
  • guiding the tractors steering as slave by the implements position as master to keep mechanics as simple as possible

Scanning the documentation and existing threads, dropping some questions in the Telegram group and here in the forum, I learned that there is little hope that any solution out of the box might fit.

Looks like I have to go the hard way to be able to help out myself.
Let’s start learning.

Along the way, there are some windfall projects and milestones:

Beyond the horizon, I consider to switch to dam farming.
One of my ideas might require to keep front and back tools in sync at ~ 2..3 cm even in curves, changing side hills and changing soil and drift conditions.
No clue yet whether at all and even if, with wich effort this might be possible.

1 Like
2

You should post all your stuff on 1 subject so the forum doesn’t get so cluttered.

2 Likes
3

Well, it’s always difficult to guess forum operators’ preferences. :man_shrugging:
But I’ll try to behave :wink:.

So where would you prefer reports from the setup of a UM980 base station?
Here in what I’d plan as my main project thread?
In the dual antenna thread, as the UM982 initiated the need?
Or not at all?

4

No, one thread, one subject is correct.

What we don’t want is someone asking the same question in 2 thread and also people going off topic of the thread subject.

But as everyone can see, there’s not much moderation here. Probably too much busy farmers not technically inclined in the moderators group :sweat_smile:

5

Curves and sidehill, this will be quite a challenge to harrow.

I think AgOpenGps v6 will not give you satisfaing results.
Begining with the fact that two different vehicles geometry will be used, the tools will be at completly different track on curves.
Also, the trailed seeder could be to far from the reference line quite often on sidehill.

But the good news is that there is an alternate version available!
I don’t know when if it will be an official release in a foreseeable future for it.

It’s the AOG_dev repository in Brians GitHub.
It’s not for the average user yet.
It’s completely different from AgOpenGps in many ways.

For you case, it will have tool steer, so you could at least have the seeded rows close from the reference line, to ease the harrowing later.

You could even add a cylinder to the harrow to compensate if the you feel the tool doesn’t follow the line closely enough.

Remember that AgOpenGps just steer the tractor axle position to the line so far, no tool steer so far, either active or passive.

Some also made their tool steer by using 2 AOG systems. One for the tractor and one for the tool.

6

Yes, that’s roughly the information I’ve collected thus far.
Actually, it were the tool-steering people that told me to get some basic AgOpenGPS experience first.

“Tool steering” in mind is the reason while already two UM982 are lingereing on my test bench.
And one of the resons to keep receivers out of the teensy box.

7

Yeah
Surely mounting an AgOpenGps system to get the hand is a good idea, especially if you never used commercial steering systems.

There are a lot of functionalities and it’s better if you can test it in less sensitive operation than seeding first!

The first basic test is just a tablet and receiver trough USB to use the system as lightbar. Then you can already drive around the field, make boundary, create lines from driving, from boundary. Create tools, and so on.

For returning into row crop it can even be enough as you find the correct row then just steer manually. I had the sprayer tractor this way for a couple of years, AOG autosteer seeded then just “lightbar”(just the tablet actually) to spray the row crop.

But autosteer is so nice, then toolsteer…
Surely you can go very deep (an fast) in this rabbit hole!

8

can’t withold sniffing :fox: there, already :man_shrugging:

There is this thread referring to some PGN that might receive nudging from some button:

When I’m not mistaken, this is the (current?) definition of this PGN
‘Boards/PGN.md at main · AgOpenGPS-Official/Boards · GitHub’

So, in standard setup,

  • AgOpenGPS on the WIN tablet sends some PGN via UDP to the teensy, to tell about desired steering angle
  • some other box may send other PGN via UDP to (where? - the teensy? the main application) to request some nudging

My idea now is to have two distinct installations of AgOpenGPS, one on the seeder and one on the tractor.
And some “man-in-the-middle-tweaker”, capable of receiving, parsing, sending PGN.
As linux-addict, I think of some perl script, maybe with the help of netcat or so.

This middle-box acts as steering for the seeder, i.e. it receives desired correction of the track.
And it acts as "nudge-button-set" to the tractor, i.e. it tells it to perform that correction.
May be, some PID loop may suffice for first trials.
Even the sidehill correction might easily be added at that point.
It’s just required to hook into the PGN-over-UDP link.

As Wikipedia tells us, PID has been developed by observing the actions of helmsmen on ships in narrow tracks. Not that different from keeping a seeder on track? And I even think that my own manual steering resembles some concepts of PID.

Early testing might be possible with lightbars, too.
The goold old track markers are just one SCV button away for assistance.