I agree
Minus 28 here this morning, so that’s a big no on testing with Ms JD AgraBot in the snow…
They kind of messed up on our weather forecast, it sure a lot colder than advertised.
Good day to spend behind the laptop.
40’s and 50’s here for the 7 day, and dry. I have a spare tablet that I use for testing. If you want me to test it on some curves and sidehill, just shoot me a link on telegram.
I’ll probably just start doing the incremental updates in github - then anyone can just do them. Not as a release but source code.
For the curious, look up Ramer Douglas Peucker algorithm Ramer–Douglas–Peucker algorithm - Wikipedia
v4 will have a simplified farmer version of it to dramatically reduce point count, increase speed for all the different boundaries
Progress so far.
I pulled it up, but don’t know how to apply it.
OR is this just to review the code ?
Just the list of changes history so far. Just download the master if you want the current source
We use PP only because I haven’t been able to get Stanley to work on our rear steered cotton pickers. We’re using contour mode mostly on circular center pivot Irrigator paddocks.
In saying that, we’re running an AOG version from about 18 months ago… Maybe Stanley has changed since then? I’m upgrading hardware/updating to V4 in the next couple of weeks.
Today I tested Stanley steer algorithm again on a flat “calmy” field, first time with dual antenna. It works pretty well, better than I remembered. So is it all about ideology? Hope that I will be able to collect experiences about the offset caused by a sidehill next days.
What is funny is they are very similar. PP uses the lookahead point as Stanley’s “steer” point.
Stanley has really no settings - but what is important is if your set point is 3 degrees for example - it needs to turn 3 degrees and have the right wheelbase. If not , it will wander back and forth.
V4 is a really good program I like it very much. There are little things I’m missing - two additional snap left/right buttons (I would like to have some for small and some for big distances at the main frame as V3 had), -the possibility to give a self defined name to a ab-line (like “next to street” or “at north side”), - and a autosave-function for a snapped ab-line (Example: I created two ab lines, snapped one of it by 1.5m and switched beetween them. The snapped distance is lost. Hope there are some supporters to that wishes.
The first sentence is very interesting. But the second one may be quite important to know for everybody struggling with stanley, isn’t it?
While you are making serious modifications to the software. I’ve been playing around with adding more data to the vec3. There are only a few locations that need modified to allow each point in the patch list to contain other data. With this data, you can set up layers and visuals to see speed moisture, and a host of other things. One in particular, is current task. And to be able to toggle between them. We loose a lot of information when we close a field. I have noted that the section files are amended and not re-written. Either, load all the information up front or read it as needed.
Here is what I did to the vec3, of course it is not just a simple three point vec. U-turn is where I found the most changes are needed. There are far better ways to do this. But here is where I started to see if it would work.
public struct vec3
{
public double easting;
public double northing;
public double heading;
public double speed;
public double temperature;
public double moisture;
public vec3(double easting, double northing, double heading)
{
this.easting = easting;
this.northing = northing;
this.heading = heading;
this.speed = 0;
this.temperature = 70;
this.moisture = 0.65;
}
public vec3(vec3 v)
{
easting = v.easting;
northing = v.northing;
heading = v.heading;
speed = v.speed;
temperature = v.temperature;
moisture =v.moisture;
}
This is what the section file looks like:
Of course, I added it to the color “section” as well,
But then it is just added to the end. Speed, temp, and humidity.
The “65,1,Default” is a task identifier that I was working on.
65,1,Default
27,151,160,0,70,0.65
65.026,38.928,0,33.2,70,0.65
73.803,34.135,0,33.2,70,0.65
65.281,39.491,0,33.2,70,0.65
74.329,35.231,0,33.2,70,0.65
65.538,40.124,0,33.2,70,0.65
74.775,36.293,0,33.2,70,0.65
65.782,40.799,0,33.2,70,0.65
75.166,37.343,0,33.2,70,0.65
I agree, being able to name lines would be extremely useful… I often need this for different tasks…
I have to be careful not to end up making v5 out of old v4 - but there are some things that just make sense and aren’t huge to add.
V5 doesn’t distinguish between lines - they are just a collection of segments to make a path - but that also means most of the program needs to be fixed/changed. Looking back at v4, it sure makes the interaction a lot easier.
rather then change a vec3 that gets used everywhere, better to make a vecSection class?
Have a look at the recPath point - much the same
I have been able to rename ab lines. Just click on the ab line title i think?
Hello,
Not a lot of experience, but for now I have used mostly PP. My first try with Stanley, I found I was very slow to go back to the line. Perhaps a bad setting of stanley.
After I remplace MMA and BNO055 by BNO085. Steering was greatly improved in PP, +/-2cm of the line, U-Turn perfect etc… But I haven’t tested Stanley since I have monted the BNO085, I should try, it shall also be improved.
Other reason why I use PP: I mount AOG on the tractor for sugar beet seeding: the tractor speed is arround 6-7km/h while doing this work. I don’t no exactly why, but I was thinking that Standley was more suitable for higher speed (more than 10km/h) and PP more suitable for speed lower than 10km/h.
So it’s a misconception ?
What I observed during my comparativ test between BNO055 and BNO085 is that BNO055 was sometime very stable, and sometime start to constantly drift:
Note: during this data record on road, BNO085 was running on Rotation Vector mode and background calibration enable. So background calibration seems be greatly improved compare to BNO055.
Math
I have made a new github branch called v4_Upgrades. It now has integral for both curve and abLine. The little spinner in the sim allows you to add some offset (side draft - only in ABLine right now) so you can see how the integral works. Be sure to set integral to about 20% in steer settings - defaults to 0%.
Is v4.4 so all new settings.
Speaking of settings - what if there was only 1 option for setting and saving everything in the whole program rather then 3 different ones? Am going to go to xml instead of text - but don’t want to do ALL the different groups code. Each saved setting group would be all your settings in one place. After all, how many setups would you really have?
https://github.com/farmerbriantee/AgOpenGPS/tree/v4_Upgrades