You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Competition this year has a lot of 3D Mapping tasks. Using an already built tool will prevent me from losing more hair.
Team is undecided between attempting ROS2 Camera-only SLAM or using a harder tool such as LIDAR. Throwing a camera underwater and seeing how it behaves is the easiest way of determining if distortion due to waterproofing and strange lighting effects due to surface perturbations actually matters for 3D mapping.
The text was updated successfully, but these errors were encountered:
To speed this along, I've gotten permission from a company I used to work at to use the VSLAM code I helped to implement while I was there. This is, of course, proprietary code, so we shouldn't go flinging it around to aggressively. That code has been mirrored to here: https://github.com/Colorado-School-of-Mines-Robotics-Club/ExternalVslam
I'm now working on tearing into that code and making a number of modifications to it, including making it run on our Ubuntu18/ROS2 Foxy platform and having it use Jetson acceleration. It currently runs purely on the CPU, which that Nano is almost certainly not equipped to handle.
The ORB feature extraction step is particularly low-hanging fruit for GPU acceleration.
Actually, I might hold off on porting/implementing everything until we've done a proof-of-concept with the unported VSLAM running on a laptop and the camera moving around in a pool.
Need to see if the thing will have enough tracking data and be tolerant to the funky underwater lighting.
Competition this year has a lot of 3D Mapping tasks. Using an already built tool will prevent me from losing more hair.
Team is undecided between attempting ROS2 Camera-only SLAM or using a harder tool such as LIDAR. Throwing a camera underwater and seeing how it behaves is the easiest way of determining if distortion due to waterproofing and strange lighting effects due to surface perturbations actually matters for 3D mapping.
The text was updated successfully, but these errors were encountered: