This is gonna be hella hard and experimental but I wanted to throw the idea out there. The idea is that we use the images from the camera and try to get the target's bounding box centered.
Not a priority unless we find out via IRL testing that our initial 3d localization estimate for the target sucks, or that the drone's self-localization sucks.
This is gonna be hella hard and experimental but I wanted to throw the idea out there. The idea is that we use the images from the camera and try to get the target's bounding box centered.
Not a priority unless we find out via IRL testing that our initial 3d localization estimate for the target sucks, or that the drone's self-localization sucks.