Q-Planning enables recursive self-improvement for robot policies: 25% to 80% success

animesh_garg · x · 2026-08-28

Q-Planning is a learning-based harness that allows large black-box robot policies to recursively self-improve. On a fine-grained task, it boosts success rates from 25% to 80% in 30 minutes (100 robot attempts) without extra human data.

Original post →

More from Embodied

Embodied channel →