Join Date: Aug 2012
Location: Vermilion Alberta Canada
Mentioned: 19 Post(s)
Quoted: 2498 Post(s)
I'm going to be really hit and miss on replies for the next month or so, a bit of R&R. Serenaded by the sound of orthopterans this morning while i type this.....
WTalen, I think your steer motor turns too slow, its got to go much faster if you want to do those kinds of ohmega turns. Pure pursuit cuts off corners with farther look ahead points, so going outside of a turn as in your video shows its a steering issue. Mine is set around 1.8 secs ahead, min distance around 3 meters, 1.5 on the offline gain. Once the tractor gets behind on steering - very evident on the ohmega turn - you're screwed, it won't catch up again. Best thing i did was use that 12v to 24v convertor, motor spins good speed and has way more power/control. If the problem exists on contour as well, your motor isn't keeping up to the requests made, it gets behind, tries harder to keep up, starts to oscillate, bad scene.
You can fudge turning sharper but making wheelbase longer as steering angle calc is highly dependent on wheelbase. But by far the most critical success/failure aspect is can the steering wheel turn fast enough. In latest version an initial stub is added on to the start of turns to allow the tractor to start turning at the right time based on the look ahead position. If the UTurn popped up but the lookahead was already well into the turn, that was a problem.
UTurns are completely generated by Dubins algorithm unless using patterns, then they are just made by the pattern.
Why turn so sharp? Unless plowing, leave a couple swaths - set skip to 2, and fill in the blanks later, the turns will be faster and easier on your equipment. It's why we have GPS.
Mounting the antenna closer to the front/farther from pivot point gives a much greater lateral error (distance off line) and will then generate faster turns because with great distance offline, a greater steer angle is made. Downside, when going around a corner then antenna fix to fix is sharper since the arc is sharper then the arc the pivot makes. No free lunch, but the lateral error greatly improves and guidance because it detects off line much faster.
When you change lookahead distance, proportional gain needs to be changed as well, the 2 work hand in hand.
WW, that is smooth and dead accurate, wow such a nice job. You have all the right ingredients, smooth GPS signal, steady heading, quick response. Perfect.
We know AgOpenGPS works using the simulator, that is the goal point for the mechanical system as well. The closer you can get to rapid steering response, accurate and consistent GPS fix, smooth and accurate heading, the more the tractor will work like the simulator - mathematical certainty.
We can make this system better and better, keep trying new things, I'm pretty confident in saying that AgOpenGPS works!