I feel that I can write a simple SLAM by myself based on the pnp principle, and it is still VIO.
Yes, you can launch it yourself.
We generalize the QR code to something ordinary, but we don't know the side length. If we know it, then we can use the pnp principle to infer the pose of the camera.
For example, can this side length be obtained from the binocular depth map? So many problems will not be solved! ! ! ! !
In this way, you can derive a SLAM! ! !
In addition, IMU can sense the change of one's own posture, it is more accurate! ! ! ! !
。