Sources
Fleet-Scale Reinforcement Learning for Generalist Robot Policies: Learning While Deploying
A new paper trending on HuggingFace (May 4) tackles a key unsolved challenge in robotics: fleet-scale post-training of a single generalist policy across diverse tasks. The approach deploys a pre-trained policy across a robot fleet, aggregating both autonomous rollouts and human interventions into a shared replay buffer for offline and online updates. Prior work focused on specialist policies; this is the first framework to maintain generalist capabilities while learning from fleet-scale deployment data.
↳ Follow the thread