Escaping the Diversity Trap in Robotic Manipulation via Anchor-Centric Adaptation

ArXi:2605.07381v1 Announce Type: cross While Vision-Language-Action (VLA) models offer broad general capabilities, deploying them on specific hardware requires real-world adaptation to bridge the embodiment gap. Since robot nstrations are costly, this adaptation must often occur under a strict data budget. In this work, we identify a critical diversity trap: the standard heuristic of "maximizing coverage" by collecting diverse, single-shot nstrations can be self-defeating due to non-vanishing estimation noise. We formalize this phenomenon as a Coverage--Density Trade-off.