$π\mathbf{R}^2$: Reactive Real-time Flow Policies
2026-07-28 • Robotics
RoboticsArtificial IntelligenceMachine Learning
AI summaryⓘ
The authors address the problem that current robot control policies using large pretrained models are slow to react to new sensory input because their actions are planned in chunks executed without adjustment. They propose πR2, a method that lets the policy respond quickly to fast-changing signals like joint positions while updating slower vision or language data less often. By adjusting how the policy predicts actions based on hardware delay, πR2 can replan much faster and improve real-time responsiveness. Tests on real robots and simulations show their method increases task success rates compared to previous approaches.
generalist manipulation policiesaction chunkinglarge pretrained backbonesreactivityperception-to-action pipelinediffusion forcingproprioceptiondenoisingrobot replanningclosed-loop control
Authors
Sungjae Park, Shubham Tulsiani
Abstract
Generalist manipulation policies increasingly take the form of action-chunking flow policies built on large pretrained backbones. Such chunks run open-loop, so the policy cannot react to sensory input arriving mid-execution, sacrificing \emph{reactivity}. Replanning more often would restore it, but the perception-to-action pipeline (a large backbone plus multiple denoising steps) is too slow: this \emph{latency} forbids frequent replanning and leaves committed actions stale, making such policies ill-suited for dynamic, closed-loop control. We present $π\mathbf{R}^2$, which makes these policies reactive and real-time while retaining large backbones, expressive multi-modal policies, and multi-action prediction. Built on the per-position noise schedule of diffusion forcing, $π\mathbf{R}^2$ contributes two ideas. First, it splits conditioning into a fast channel (proprioception, fresh every tick) and an asynchronously updated slow channel (vision-language features), so the policy reacts to proprioception within a chunk while tolerating stale vision. Second, a latency-adaptive flow schedule treats in-flight actions as inpainting conditioning and emits actions in one denoising step per call, letting one trained model adapt to varying hardware latency. Requiring minimal modification to existing architectures, $π\mathbf{R}^2$ can be finetuned from a pretrained policy: applied to GR00T-N1.7 on a real xArm6+XHand platform, it replans closed-loop roughly $4\times$ faster than the base policy (~$25$Hz on an A5000 GPU), acting on a fresh observation every $40$ms. Across simulation and real-world manipulation tasks, $π\mathbf{R}^2$ improves the success rate by up to $23\%$ in simulation and $30\%$ in the real world over the strongest baseline. Project page: https://pi-r2-flow.github.io/