Black holes im Garten [solved]

[RoboBot]

Hallo liebe Ardumowers,

ich habe meinen Arctic Hare auf RTK umgerüstet. Es kommt Sapos NRW via WiFi zum Einsatz.

Wenn ich nach erhalt des FIX im Relative Position Mode den Perimeter und die Exklusionen zeichne, die errechneten Wegpunkte hochlade und auf Start drücke, fährt er den Rasen trotz "topologischer Herausforderungen" perfekt nach Plan ab. (Hut ab, auf den Sunray-Autoren - sehr beeindruckend soweit

Sobald ich aber in den Absolute Mode wechsele, eine Position auf dem Grundstück vorgebe, den Perimeter sowie die Exclusionen erneut abfahre bekomme ich ca. in der Mitte der Fläche das Problem, dass der Mäher über ein paar m² hinweg seine eigene Position nicht mehr aktualisiert, wohl aber die Ausrichtung nach IMU. Dieses "Loch" verschiebt sich um einige meter, wenn ich eine Andere Bezugskoordinate vorgebe.

Ich habe die Weltkoordinate auch unterschiedlichst plaziert - außerhalb, wie innerhalb des Mähbereichs. Auch mal so, dass nur positive, und mal nur negative Bezugswerte vom Due errechnet wurden - erfolglos.

Von FW 1.0.32 bis 1.0.46 das selbe Bild.

Ein Wechsel zurück in den Relative Mode samt Neuanlernen und der Schneehase rennt wie auf Schienen klaglos durchs Gelände - auch mehrmals hintereinander. Jedenfalls bis zum Ausschalten.

Nun muss ich doch um Hilfe bitten hoffe auf Lösungsansätze von den Profis hier

EDIT: noch ein paar Bilder als Orientierungshilfe...

Der blaue Bereich wird von Sunray ignoriert, aber nur im Absolute Mode:


Die Patchantenne, das Kabel ist auf ca. 40cm gekürzt.
Sie sitzt auf einer ca. 99mm x 66mm Cu-Platte welche mit der Batteriemasse verbunden ist.


Hier ist die elektronische Baugruppe zu sehen, das RTK-Modul sitzt huckepack auf dem Due.


Und noch mal aus der Perspektive, das Bild ist etwas älter.


Es vergehen ca. 30-40 sek. bis zum ersten Fix.

EndOfEdit.

Güße aus dem Bergischen Lande...

Roberto

 

[raess]

Hey Roberto!
Can you please share your map?

RF

 

[RoboBot]

Hi raess,

so, I've created three maps today and shared them by the id: 4884.

The first one shows a section, drawn in absolute position mode. Whereby I was not able to define the exclusion patches because, with the real mower entering the blind section, the virtual mower icon on the app's map stops moving forrward. As mentioned, only heading by the IMU is updated in the map. As the mower moves forward, it seems not to change it's position until it leaves the blind section physically.



The second one was made using relative mode, where everything went as expected and the mower navigated precisely through the map. Btw. exclusions 1 and 2 are located within the blind section in absolute mode.



For the third one, again in absolute mode, I drove the mower towards the blind section's center. By crossing the blind section border, the virtual mower stoped here and I was able to set an exclusion point. Then I reversed the mower in order to leave the blind section and reenter it again one or two meters right relative to the last position marked. By doing so, I could "circumnavigate" an exclusion section, whitch now shows quite exacliy the outlines of the blind section.



But the blind section's location changes by a meter or two as I change the absolute world coordinate.

Waypoints located within the blind section became overrun by the mower. Then, as the mower leaves the blind section, it becomes aware of having missed out it's waypoint. The mower turns around to catch it up. But again, the mower runs over the waypoint until it leaves the blind section, then turning around and so on. A endless loop has begun.

Exclusions, witch were hacked in by hand within the blind section, can't also be recognized by the mower. Furthermore the obstacle detection won't kick in.
When the mower rides of of the map though a blind section and "wakes up" outside of the map's boundaries it goes into kidnapped mode.

Greetz
...Roberto

ps: raess, Arctic Hare is your design, right? I love it and it was fun to print and build. Thank you for this great design!

 

[AlexanderG]

Hello Roberto,
The ublox receiver sends two positions to the Ardumower:
a) relative to the base (east/north) in meters (this is used in relative mode) - because virtual base station (VRS) coordinates move each start this mode cannot be used for SAPOS/VRS
b) latitude/longitude (this is used in absolute mode and the base coordinate set in the app is always substracted from this to calculate again east/north in meters)
I'm not describing a solution here, I will describe possible errors:
1. The GPS receiver maybe has been configured with a filter or something for this specific area (latitude/longitude) - You could use ublox-uCenter and try running it in that specific area - if the coordinates (longitude/latitude) do not update, you probably have some filtering going on there...
2. However, if the GPS receiver correctly updates longitude/latitude in that specific area, there's probably a bug in the Sunray firmware for this specific area (latitude/longitude) - The code for subtracting is in helper.cpp (function distanceLL) - maybe someone can have a look and maybe some clever eyes can see any problematic line in there... If you think the error is in the Sunray firmware, you can send me the center latitude/longitude of the specific area (and please also your base lat/lon coordinate you use in the app), and I will try out the code to see if the code behaves correctly and if not, I'll fix it.
Regards,
Alexander

 

[Eddi]

Hallo,
so wie ich die erste Nachricht von Roberto lese, funktioniert der Mower bei relativer Positionierung problemlos und macht nur im absoluten Modus Probleme.
Ich würde deshalb nicht davon ausgehen, dass es ein Filterproblem ist, sondern eher ein Bug in der Sunray Firmware. Ich schaue später mal rein, vielleicht sehe ich etwas.

Gruß

edit: okay, sorry. Hab nicht richtig gelesen.

 

[RoboBot]

Hi all!

As Eddi wrote, using relative mode, everything works fine until I shut off the mower.
Therefore I dont think to have set some filters in the RTK's configs.

Right now, I tried U-Center to get a fixed coordinate in the center of my blind section, witch worked out fine. Within 40 sec's it got his fix.

Alexander, I'll post the requested coordinates to you.

Greetz...
...Roberto

 

[RoboBot]

Hi all!

Last Saturday Alexander came over to get some measurements on site and we had a nice talk about what it's all about that Ardumower-thing
One day later, with a new Sunray build, the problem was gone.

All trouble was caused by the FiFo Buffer of the Arduino witch had to be increased.

Now the mower navigates flawlessly through the garden and all blind areas are gone.

...Roberto