My basic problem is that the mission planner does not track the rover correctly. When I drive North to South its fine. But when I drive East to West, the mission planner shows the rover going from West to East.
I’ve seen this problem as well in the past and it was caused by having the beacons in the wrong order. You’ve put the beacons in the “N” shaped pattern shown on this page? http://ardupilot.org/copter/docs/common-pozyx.html
Oh Yes, the anchors are arranged in the correct order but I have made several alterations to your system described on that page.
The Anchors are not in a perfect rectangle… they’re in a trapezoid form.
I’ve disabled the txpower setting function. Primarily because it just won’t work consistently.
I’ve set the ‘anchors to anchors’ distance reading to actual distances rather than letting the sketch do the get.ranging function… again because of inconsistency.
Number 2 & 3 above were made as a result of communications with the POZYX people.
Sometimes when I fiddle with the BCN_ORIENT_YAW setting, I get the East West to work but the North South is backwards.
Another issue I have is while manually driving a short distance, the tracked path (on the mission planner) overshoots the stopping point then eventually comes back. I assume this is the EKF doing its job. I’d sure like to reduce the EKF influence but that is way above my pay grade.
So are there others out there using POZYX on a rover and if so what are your experiences. And is the 3.2 section the best place to share those experiences?
There are at least two other people who also want to get Pozyx running with Rover so I think this is the place to discuss it. Rover-3.2.0 is certainly the version to use.
Sounds like you’re confident fixing up the sketch, so I guess worst case, you could just reverse the sign of the values for one axis. It shouldn’t be necessary but that’s all that should be required I think.
OK I’ll try that. Quick follow up. It appears you send 1) distances from the each anchor to the tag AND 2) the x&y do.positioning values to the Pixhawk. Why both sets? With that info then I could take a better guess as to which value I want to turn negative.
I changed the msg.info.y = anchors_y[i] to -anchors_y[i], then tweeked the Yaw to 90 degrees CW more than it should be… and now it tracks pretty close… BUT…
I sure would like to know more about how the Arduino sketch and the MP BCN variables affects the MP tracking… cuz my robot’s actual location is not yet close enough.
Since the POZYX send out pretty accurate data, how do I reduce the EKF involvement with tracking calculations. I use Megunolink to plot my actual POZYX data and its pretty consistent.