Fetching the paper…

VANP: Learning Where to See for Navigation with Self-Supervised Vision-Action Pre-Training · Around