I’m having trouble figuring out turns in Auto. The aircraft always just goes right to the Lim_roll_cd, regardless of where the next WP is. Is there any logic about shallowing out wider turns?
It also seems to always overshoot, despite having a WP radius of 200m, the aircraft always drives into about 110m to make a 90deg turn at 26m/s and always overshoots (lim_roll cd = 2500 cdeg) is there a way to tune this?
I would also like the ability to turn off xtrack correction, sometimes I want to maximize my straight leg in a small field, and end up spending half of the length correcting back to course after overshooting (see above)
There is also no apparent wind correction, steepening turns with a tailwind and shallowing going into a headwind, turing early with a cross from inside the turn, or turning late with a crosswind from outside the turn.
Have you read about tuning the navigation (L1) controller? http://ardupilot.org/plane/docs/navigation-tuning.html
I saw it at one point but haven’t fooled with it much. In SITL turning up (increasing) the NAV period does make the aircraft turn earlier, but slightly wider, which makes sense. Although, SITL also has something seemingly messed up in its native turn simulation as level turns at a given bank always overshoot the actual turn radius that should be expected at that bank (ie. a no wind level turn at 22 m/s and 20deg AOB will always have a radius of 136m, but in SITL its more like a 200m radius actually flown). I will try this with the real aircraft, which does obey physics, and scribes a 136m turn at 22m/s and 20deg AOB, and see how it does. Thanks!
At 26m/s and 25 degrees of bank, the best turn diameter is about 300m. That’s a lot. Are your flight lines separated by at least 200m? Can you please provide a log or a screenshot of your ground track?
Is there a reason you can only do 25 degrees of bank? With only 20% more load on the wings, you can reduce your turn radius by approximately 50% (40-degrees of roll).
Yup, its a lot, but its what we need right now unfortunately. we are currently min running our thrust/weight and that load factor keeps us safe. In the end for our use case we won’t be executng a lot of tight turns so the turn radius is also not a big deal there. This was more about me getting frustrated over not being able to have the flight path look like I think it should. I think the L1_nav_period is the answer.
Yes, the period affects the turns a lot.
Well, I had my thinking backwards, so had previosly turned the period down, and things “got worse”. I abandoned investigating at that point to tackle more relavant stuff, but now I see where I went wrong before, we actually need to turn it up to see what we want.
There definitely is a balance, and turning up or down your bank angle affects the tuning of the L1 control period. Changing my aircraft from 60-degrees to 50-degrees caused navigation oscillations that required a higher L1 period. The higher the bank angle, the faster the turn, and the faster the period must be to navigate precisely.