Abstract

Vision-Language-Action (VLA) models enable robots to understand and perform complex tasks from multimodal input. Although recent work explores using reinforcement learning (RL) to automate the laborious data collection process in scaling supervised fine-tuning (SFT), applying large-scale RL to flow-based VLAs (e.g., π0, π0.5) remains challenging due to intractable action log-likelihoods from iterative denoising. We address this challenge with πRL, an open-source framework for training flow-based VLAs in parallel simulation. πRL implements two RL algorithms: (1) FlowNoise models the denoising process as a discrete-time MDP with a learnable noise network for exact log-likelihood computation. (2) Flow-SDE integrates denoising with agent-environment interaction, formulating a two-layer MDP that employs ODE-to-SDE conversion for efficient RL exploration. We evaluate πRL on LIBERO and ManiSkill benchmarks. On LIBERO, πRL boosts few-shot SFT models π0 and π0.5 from 57.6% to 97.6% and from 77.1% to 98.3%, respectively. In ManiSkill, we train πRL in 320 parallel environments, improving π0 from 41.6% to 85.7% and π0.5 from 40.0% to 84.8% across 4352 pick-and-place tasks, demonstrating scalable multitask RL under heterogeneous simulation. Overall, πRL achieves significant performance gains and stronger generalization over SFT-models, validating the effectiveness of online RL for flow-based VLAs.