Why not incorporate LiDAR data directly into the Vision Transformer, just like camera input, for the rover’s navigation system?
That’s totally plausible. TRanformers really don’t care about the input sensor type. You just have to know how to tokenize correctly an efficiently. The only issue is if the TRanformer model has some specific architecture only compatible with images an even so, you can make some kind of transformation of laser data to an image version of it.
This topic was automatically closed 5 days after the last reply. New replies are no longer allowed.