Patch Policy: Efficient Embodied Control via Dense Visual Representations

Pretrained dense visual features from Vision Transformers (ViTs) are powerful yet have been underutilized in robot learning.

Pretrained dense visual features from Vision Transformers (ViTs) are powerful yet have been underutilized in robot learning. Modern robot policies either compress each observation into a single global token, or rely on visual backbones trained from scratch, sacrificing both…