I’d suggest it should swap to vision based feature tracking, or better, already be using vision-nav for position verification and heading in the first place. I suspect the core does NOT use it because the S1 was the first module out the door, and there’s no such thing as “seeing anything” when you use a snowblower, lol. The S1 is why the twin GPS technique was created, for heading determination, and it works well for that. It is not optimal for the SAM, M1, or B1 however.
Luba2 uses vision extremely well, with just one pair of fisheyes that are forward facing. Yarbo’s vision (fore and aft) is more than adequate to knock that robot right out of the park, once the Dev team has some time to put into it.
That’s the idea – to get Vision going and use a combination of technologies to navigate around.
Yarbo is looking into VSLAM Fusion, which I think it’s called (Bryan knows this stuff – I just parrot his brilliance to make myself look good). I thought that was a ways out, though.
Have a CLOSE look at the new mower, which I brightened up as I could with the editor I was using (which isn’t much).
Some kind of small dome on top, seemingly square, flat front section between stereo fish eyes…? I’m not sure I smell LIDAR, but maybe an enhanced camera system to make Vision easier to refine? (OK, maybe I smell LIDAR. Should I smell LIDAR?)
My thing is what does this NEW idea mean for everyone that already has one? A new one, that haven’t even got to use the one they have due to. Well some know my story, that hasn’t even got fixed yet. Wait and see? Offer a updated newer one for everyone that is already on board?
I too have spotty GPS signal where I have currently installed the charging station. I gather from this thread that the current logic is for the rover to check one time before the scheduled start time for GPS. It skips the schedule if GPS is not adequate. Why doesn’t it check periodically between the defined start and stop times? For e.g. it could check every 30/60 minutes for GPS to improve between the start and stop times. The one-time check needs to change. Yarbo should not expect that a GPS signal is available at all times while parked.
