Q-Planning Enables Self-Improving Robot Policies via Small Q-Functions

Researchers led by Animesh Garg introduced Q-Planning, which pairs frozen behavior-cloning policies with a small off-policy Q-function for self-improvement from successes and failures, cutting human intervention and hitting 99% on LIBERO-10.

2026-09-17 ~ 2026-09-18 · 3 related posts

• Q-Planning Enables Self-Improving Robot Policies: LIBERO-10 Hits 99% With Frozen BC — chris_j_paxton · 2026-09-17
• Q-Planning: frozen BC policy plus small Q-function enables robot self-improvement — animesh_garg · 2026-09-18
• Q-Planning adds a Q-function to policies like pi-0.5 so robots improve without interventions — micoolcho · 2026-09-18