ProgressVLA: Progress-Guided Diffusion Policy for Vision-Language Robotic Manipulation

TL;DR

ProgressVLA introduces a novel progress estimation and guidance framework for vision-language robotic manipulation, significantly improving success and generalization in long-horizon tasks.

Contribution

The paper presents a robust progress estimator trained on large-scale datasets and a differentiable progress guidance method using an inverse dynamics world model.

Findings

Achieves low prediction residual of 0.07 in simulation

Demonstrates zero-shot generalization to real-world samples

Improves success rates and generalization on benchmarks and real robots

Abstract

Most existing vision-language-action (VLA) models for robotic manipulation lack progress awareness, typically relying on hand-crafted heuristics for task termination. This limitation is particularly severe in long-horizon tasks involving cascaded sub-goals. In this work, we investigate the estimation and integration of task progress, proposing a novel model named {\textbf \vla}. Our technical contributions are twofold: (1) \emph{robust progress estimation}: We pre-train a progress estimator on large-scale, unsupervised video-text robotic datasets. This estimator achieves a low prediction residual (0.07 on a scale of $[0, 1]$) in simulation and demonstrates zero-shot generalization to unseen real-world samples, and (2) \emph{differentiable progress guidance}: We introduce an inverse dynamics world model that maps predicted action tokens into future latent visual states. These latents are then processed by the progress estimator; by applying a maximal progress regularization, we establish a differentiable pipeline that provides progress-piloted guidance to refine action tokens. Extensive experiments on the CALVIN and LIBERO benchmarks, alongside real-world robot deployment, consistently demonstrate substantial improvements in success rates and generalization over strong baselines.