I shelved her after a crash and a thousand dollar scare. I moved on to more comfortable projects that don't defy gravity.
I could leave it shelved, intact and safe. But no, titan dynamics said I could use their airframe in my dissertation. I paid $200 in tariffs for monster ESCs and brushless motors. It's time to crash her or add the coolest flourish I can imagine to my defense, ego, and resume.
New motors are gonna take fresh mounts on the arms and re-leveling which was a pita. I'm ordering a gh1.25 pre-crimped because I'm so afraid of another short in the same place. I picked up some 6Ss which should get us to 2x thrust to weight ratio.
Should be done before the weather is nice enough to fly. Then hover tests and quad calibrations. After that, I guess you just punch it?
I'm so scare to try flying again. ora pro me
@BecomingCritter it only takes one or two virtually photorealistic engineering marvels collecting dust on your flat surfaces before you‘re completely dulled to these things which are undeniably remarkable
@yacineMTB They put out SAM and DINO. Their vision work is insane.
They vision work is all in service of their AR products, but the localization and perception works on robots too.
@_Stocko_ Makes total sense. I started out looking to do something similar and then talked myself down to a regular (easier) balancing brick and then just talked myself down further to just focus on SLAM for a treaded bot lol
Going forward I don’t really see myself using the OAK depth online much if at all. LIO is very very good and I’ll have LiDAR slam finished next time I SSH into this thing.
I could definitely see fusing them for an extrinsics calibration.
I’m toying with putting SAM or YOLO on the TPU instead to label lidar points.
@SanSanychUA Hi. Full disclosure, I thought I was going to get reliable VIO from just the OAK-D but I really really didn’t. I think part of my trouble is that the OAK-D is only a couple inches above the ground and half the FOV (floor) is undepthable. This may have been a skill issue.
@moaazsidat@comingconflict I strongly considered the RPLidar. Have you been happy with it?
Are you doing slam on these bad boys? SLAM integration, even VO seems so much more difficult on outdoor bots than just GPS/RTK