Just rebooted and got the V3.9.16 firmware (the hotfix for “rover stops while working”).
I started a Work Plan that has a few Deadends. When the rover is looking to go to and start the next Deadend, it sits for a while (20 seconds? 30? guessing?) doing nothing. It will eventually start up and go mow the Deadend. Next Deadend, same thing. The Work Plan is now on the zigzags, and those are going OK.
This did not happen when I ran my first Work Plan today which also has Deadends lined up. This was with FW V3.9.10.
The Diagnostics page didn’t show anything out of the ordinary.
Yes. I had rebooted and got the new FW and sent it out, but didn’t re-reboot right then.
After noticing the hesitations, I rebooted with hard power cycles on everything - rover, WiFi Extender, Data Center PoE. When I resumed the Work Plan, it hesitated, so that could point at my crazy map or something in the way the Work Plan was calculated originally, perhaps (I didn’t restart the Work Plan – I resumed it).
There’s add’l detail (I couldn’t edit the post):
After the hard power cycle, it hesitated for another Deadend or three, but then I think it mowed a Deadend twice (I wasn’t watching closely, unfortunately).
Then it did the zigzag, no problems.
It mowed another Deadend after the zigzag and did not hesitate, but it mowed that Deadend twice for sure (I know it because it did so right in front of me to impress me or something).
The final Deadend of the Work Plan went OK – no hesitation, mowed once.
I mowed another Area after that with several Deadends, and there were no delays and no repeats.
So along the way it got slightly weirder but then started sorting itself out.
I’ll send the rover out to that same Area today to see if it’s reproducible. I smell a one-off, but I’ll check.
If Yarbo peeps are curious and want to poke around - MowFrontRight+DEs, start time 2025-07-04 14:32 US EDT.
Started a Work Plan on that Area again. There is still a slight hesitation when it enters the Area from the Pathway, but it’s not NEARLY as long as before. It went to the Deadend, hesitated that little bit again, mowed the Deadend. So it’s better.
I sent it to mow another Area, and it hesitated that small bit existing the Pathway again. Else it’s mowing OK. There are no Deadends in that Area, so I can’t check on the repeats.
OK, I’m not seeing the LONG hesitation now. I will keep an eye out for the Repeated Deadend Syndrome.
Anecdotal datapoint - I may have had a DE repeat back on .8.18. Only happened once, and I wasn’t sure it actually did it. But this might go back to at least that FW.
I’ve seen it do a DE several times in a row if obstacle avoidance / bumper hits happen at the end of the lane, so I wonder if it could be related to vision artifacts.
@Steve@Ken just to be certain. When you created those particular DE’s, did you go straight in and then stop the recording? OR, did you potentially also back up and then stop the recording? Might explain the double effort.
I’ve had the issue with a dead end getting mowed twice. I had to recreate those. I had another that would just get skipped entirely. Had to recreate that one too
@bryan.wheeler - Start, go to end, Next. No double-dippin’. These Deadends have all worked properly in the past.
The one that definitely did a double last night, @steve, has a small post along the way that it hooks around, but nothing at the end or the start. It’s been fine - I don’t think the vision system is in a panic.
I need to double check the video on this one from today: The mower seemed to handle the parallel Deadends I have in the Dog Pen differently. Is it me, or does the rover seem to be going out of its way now to navigate stupidly, even worse in V3.9.16? Rocking back and forth, turning around only to turn back in the direction is was already pointing, having a hell of a time just getting to the border around No-go Zones… Maybe it’s me. The rover should be leaving as tiny of a footprint as it can.
Oh, it mowed the first Deadend in the Dog Pen in reverse. It’s not done that before. Yes, the end of the Pathway is near the start of the Deadend, so technically this isn’t “wrong,” but the behavior was different today.
Well, NOW Bluestacks no longer works from my laptop as it was before. I thought Bluestacks used the laptop’s WiFi and the Data Center routed that via HaLow. Same with the phone? Not now anyway – either. I have another post waiting approval for this discussion about my seeing 4G all day (I recently enabled it, but it’s now off again) – we can pick up this discussion there. Maybe I just don’t understand how this works.
Anyway, a few Work Plans run today, they all mowed and finished, just need to look at the video from the Dog Pen with those Deadends.
Regarding the issue of the rover mowing Deadends more than once, I’ve passed this to the relevant team for further investigation. Your feedback is greatly appreciated and helps us improve.
Took a quick look, and between V3.9.10 and V3.9.16 there was a slight behavior change with my stacked Deadends in the Dog Pen. In the later firmware Yarbo came out of the 3rd Deadend and spun too far, missing its next Deadend, then it had to correct by spinning more and rocking. It then did the final Deadend but a little off the mapped path and didn’t go quite to the end.
So, yes, I did see something, but this seems more like a slight GPS error than navigation. If I ran it again today, it’d probably do what it did before.
