Perhaps I trust AOG math better than AOG does… the picture looks great! 
Understood. I also would assume that there are quite a few failure modes that would have to be explored as well.
Word choice should have been “calculates closely” instead of “knows…”
For a simple approximation, it does rather well. I think I posted a long time ago with a description of how the algorithm works. My hunch is it tracks a little tighter (closer in on a turn) than the real thing does.
Definitely it does not take into account any kind of side slip, or if the implement doesn’t pull straight.
I suppose that instead of 2 feet off the curve it may be able to get 4-6" off the curve?
6 row corn planter, 12 row head… maybe there are other solutions… perhaps a 24 row planter… 
But why does the implement need to follow the wayline closely on a curve in your case? It does about the same distance on the next path and then coverage is almost perfect. Just record the contour path when you drive the first pass taking into account how the implement drives on curves.
Even if I mentioned an example where it makes a difference if the implement follows the wayline, I’d say most often for single receiver systems it does not matter if it is the implement or the tractor following the wayline. Any more examples where it is important?
Now if the implement needs to follow the wayline accurately, I’d accept a second receiver approach if it makes the SW any simpler. Our RTK receivers anyway are so cheap.
Using an 8m cameleon rowseeder (25 cm pr row of barley or beans)
It is towed so obviously at sidehill it wil be maybe 15 cm beside tractor track, not a problem in distance between tracks as sidedraft is almost same track to track.
Problem occur when you go cultivating between rows( same cameleon which has camera side adjustment (+÷ 15 cm) of frame, to stay correct between rows.
But on sidehill tractor must go uphill and step on rows to keep the dragged machine enough uphill to be able to stay at correct position between rows.
If machine had been steered by another GPS antenna, then machine could have made rows at correct line , and also steer machine uphill when cultivating.
You can even buy a cameleon with hydraulic steering of carrying wheels, and if you have a sbg autosteer system with dual line two antenna steering, you are good to go 
I see raven has now taken over sbg.
This is a very valid scenario for the implement antenna option. But I thought the discussion in this thread was about AOG (or any other auto-steer system) with a single antenna on the tractor but estimating the implement path and trying to drive the tractor to keep the implement on the wayline, not the tractor. This approach cannot compensate sidehill drift for the implement.
Raven still offers implement guidance. It’s over ISO-Bus now and comes up as an implement in the universal terminal.
We used it with plough guidance. It’s definitely a different approach as what SBG had, but if you are used to it it is much easier, mostly simpler to install because all the calibration happens in the UT, and everything is over the bus.
I agree that implement guidance is the best solution for these problems. There are some situations where the simpler calculated solution could improve enough that implement guidance would not be necessary, but could be improved by guiding to the calculated point of the implement.
The other situation that I have is strip till and planter. Sidehill drift is not a big problem as it is fairly closely matched by strip till and planter. However, when following a curve, there is typically enough difference in the track of the soil warrior to the track of the planter that it ends up being “no-tilled” in some places. Yield associated with it probably isn’t enough to justify implement steering on the planter, but would be nice to explore what we are talking about.
The primary use I can see for implement guidance is in situations where one implement needs to follow the track of a pass made earlier in the season. There are two different types of guidance that are needed.
1 - (Active Implement Guidance) In this scenario the implement and tractor both need to stay on the same track. This requires some hardware to correct the implement for error. The most common hardware is either steerable coulter’s on the implement or a movable hitch on the tractor. This approach would require an additional receiver as well as hardware to steer the implement, so the cost would be significant.
2 - (Passive Implement Guidance) In this situation, only the implement needs to stay on track, so instead of hardware on the implement, the tractor steers OFF track to keep the implement ON track. This approach would only require an additional receiver, so it should be a cheaper option, but it would require some significant software upgrades to AOG.
In my opinion, with the low cost of gnss receivers, it would not be worth spending a lot of time on theoretical calculations only to end up with mediocre results.
Thanks for the summary.
I would like to understand one additional item related to the passive implement guidance.
To steer the tractor off track to put the implement on track requires a position error that could be provided in two ways
1 - From gnss receiver mounted on implement. (Would this “zeroing” need to be done regularly if not on RTK?)
2 - From calculated position of implement. Don’t we already have this with the geometry and trailing algorithms that are in AOG with a reasonably accurate location? Position error is already available from calculated implement point to line? Substitute this point into the guidance to calculate a correction in the steering.
-Use the difference in positions to add appropriately to the geometry of the path and tune corrections?
-Perhaps switch back to standard during headland turns.
-look ahead leading implement guide point, not leading tractor guide point
I’m way over my head here, probably over-simplifying, and will defer to the experts 
I am planning on working on the passive steering system this next winter. I will work on it using both actual GPS data and using the calculated tool point in AOG. This is a project that I plan on taking on myself if someone else does not. This last year I have been getting AOG on several pieces of equipment on our farm. This next year will be doing testing and improvements to what I have now.
Our farm is mostly farmed on the curve all of our fields have terraces that we follow when planting. Each pass will have more and more skip if the implement does not follow the curve.
I went to a John Deere technology day a few weeks ago. They are doing passive implement steering with GPS on the implement now. This is to get as accurate a gap between passes so a 18 row corn planter could be used with a 12 row corn head. Or a 12 row planter could be used with an 8 row head.
Very interesting but I just don’t understand why each pass would have more and more skip if the control point was at the tractor? AOG would use repeating contour lines, not an adaptive line based on the previous pass (coverage). The adaptive line should have more skip/overlap but even there I guess not increasing from pass to pass. I’m probably missing something here.
The issue is that the AOG lines are parallel to each other. This is all fine and good. That is actually where you want the implement to go is parallel to the previous pass. The problem is that when the tractor goes parallel the implement will not due to the physics of dragging an implement behind the tractor. The tractor may be within 1inch of the guidance line but the implement may be a 12" or more off the line depending on the size of the implement and how tight a radius the curve. The tighter the curve the worse the overlap or skip. This curve radius changes pass to pass. Some will get tighter while others will get larger.
When manually steering these lines you will intentionally drive past the guidance line on the curve to pull the implement onto the curve. The same would happen with GPS. The tractor position would not be kept on the line. The tractor GPS would only be used to determine the steer angle, or the angle of the tractor relative to the implement (calculated WAS). Then the steering algorithm would steer the tractor to keep the implement on the line. They will not both (tractor and implement) be on the line at all times. The implement will track the line creating parallel passes.
I will try to get some video footage this spring when I am drilling beans. Since the tractor will be steering itself I can have time to take videos. ; )
For question 1, if not using rtk there is no purpose in attempting implement guidance. Implement guidance is to get the last inch of accuracy possible. Without RTK the position measurement will be too unstable to help.
Implement guidance in long straight rows is for the last inch. I think in curvy East Ohio fields it can correct feet in some instances… maybe…
By the way, thanks!
Amazes me how many of these gears have popped up. 3D Printing is a cool tool.
This sounds great! I’ve been waiting and hoping that AOG will get this capability someday. I’m in Southwest Iowa, and we farm terraces as well, so I would love to be able to plant directly on top of my fall NH3 strips. I’ll be waiting to see what happens next winter!
If your lucky enough to have flat land growing cerals another strategy to line up seed rows or get inter row effect with less accuracy is using a slight offset.
So if last years stubble is planted at 90 degrees, try the next at 92 degrees. You will cross over rows but most will be planted between the old rows. Really neat look from the combine seat the second year.
Hello, I am new to AGO, but also thought about this issue playing around on the sim while waiting for my PCBs. E.g. the change of control point between PP (rear axle) and stanley controller (front axle) is clearly noticeable. Implementing some sort of kinematic tractor-trailer model controller seems the right way to approach this and there are numerous papers on the topic.
Thinking about side slip on sidehills: as a first try a roll-dependent offset on the tractor/trailer angle could be used thats tuned through the UI.
As we are using a single-sided rear mounted mower a kinematic controller would also be interesting for its unintuitive path (when steering left the mower first moves right, and behavior between left and right turns is vastly different)
Wonder how much of an actual improvement it would yield though, since some assumtions/approximations will be made… Would be happy to discuss / contribute
i hope it works out for you! if you get an implement receiver on the implement and get the tractor to keep implement online im all for it . go for it!
