In my recent work on Multiformer, I explored the power of lightweight hierarchical vision transformers to efficiently perform simultaneous learning and inference on multiple computer vision tasks essential for robotic perception. This “shared trunk” concept of a common bac...