GR-RL: Going Dexterous and Precise for Long-Horizon Robotic Manipulation

TL;DR

GR-RL is a novel robotic learning framework that enhances generalist vision-language-action policies into specialized, high-precision manipulators capable of complex, long-horizon tasks like shoe-lacing, by filtering demonstrations and using reinforcement learning.

Contribution

It introduces a multi-stage training pipeline with demonstration filtering, augmentation, and reinforcement learning to improve dexterous manipulation performance.

Findings

Achieved 83.3% success in shoe-lacing task

Demonstrated improved generalization through morphological symmetry augmentation

Showed effectiveness of offline RL with sparse rewards for progress estimation

Abstract

We present GR-RL, a robotic learning framework that turns a generalist vision-language-action (VLA) policy into a highly capable specialist for long-horizon dexterous manipulation. Assuming the optimality of human demonstrations is core to existing VLA policies. However, we claim that in highly dexterous and precise manipulation tasks, human demonstrations are noisy and suboptimal. GR-RL proposes a multi-stage training pipeline that filters, augments, and reinforces the demonstrations by reinforcement learning. First, GR-RL learns a vision-language-conditioned task progress, filters the demonstration trajectories, and only keeps the transitions that contribute positively to the progress. Specifically, we show that by directly applying offline RL with sparse reward, the resulting $Q$-values can be treated as a robust progress function. Next, we introduce morphological symmetry augmentation that greatly improves the generalization and performance of GR-RL. Lastly, to better align the VLA policy with its deployment behaviors for high-precision control, we perform online RL by learning a latent space noise predictor. With this pipeline, GR-RL is, to our knowledge, the first learning-based policy that can autonomously lace up a shoe by threading shoelaces through multiple eyelets with an 83.3% success rate, a task requiring long-horizon reasoning, millimeter-level precision, and compliant soft-body interaction. We hope GR-RL provides a step toward enabling generalist robot foundation models to specialize into reliable real-world experts.