Berkeley paper: first tractable estimator of empowerment from offline data for exploration
svlevine · x · 2026-10-21
Sergey Levine's team (UC Berkeley/Princeton) introduces the first method to estimate empowerment—an agent's measure of control over future states—purely from unlabeled offline data, without environment interaction. By jointly learning an empowerment-maximizing policy with its discounted state-occupancy measure, the estimator trains on trajectories from any unknown policy. Empowerment-high states need more exploration, so the pretrained estimator improves online goal-conditioned RL exploration. Code released.
More from Embodied
- SF hard-tech crowd hosts real humanoid robot fight night — edgarpavlovsky · 2026-10-21
- REK robot fighting league footage goes viral: Terminator-style mechs battle in the cage — edgarpavlovsky · 2026-10-21
- Hot take: home automation wins by standardizing spaces, not building robots — alysha_lobo · 2026-10-21
- CFG lifts single-step discrete action success from 27.1% to 49.2% in robot policy training — YouJiacheng · 2026-10-21
- Menlo Research open-sources Cyclotron, an RL training framework for its Asimov humanoid robot — freelerobot · 2026-10-21
- NVIDIA demos interactive AI surgical simulator running on a single RTX PRO 6000 GPU — Scobleizer · 2026-10-21