PEEK: Guiding and Minimal Image Representations for Zero-Shot Generalization of Robot Manipulation Policies

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Zhang, Jesse, Memmel, Marius, Kim, Kevin, Fox, Dieter, Thomason, Jesse, Ramos, Fabio, Bıyık, Erdem, Gupta, Abhishek, Li, Anqi
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866912600180129792
author Zhang, Jesse
Memmel, Marius
Kim, Kevin
Fox, Dieter
Thomason, Jesse
Ramos, Fabio
Bıyık, Erdem
Gupta, Abhishek
Li, Anqi
author_facet Zhang, Jesse
Memmel, Marius
Kim, Kevin
Fox, Dieter
Thomason, Jesse
Ramos, Fabio
Bıyık, Erdem
Gupta, Abhishek
Li, Anqi
contents Robotic manipulation policies often fail to generalize because they must simultaneously learn where to attend, what actions to take, and how to execute them. We argue that high-level reasoning about where and what can be offloaded to vision-language models (VLMs), leaving policies to specialize in how to act. We present PEEK (Policy-agnostic Extraction of Essential Keypoints), which fine-tunes VLMs to predict a unified point-based intermediate representation: 1. end-effector paths specifying what actions to take, and 2. task-relevant masks indicating where to focus. These annotations are directly overlaid onto robot observations, making the representation policy-agnostic and transferable across architectures. To enable scalable training, we introduce an automatic annotation pipeline, generating labeled data across 20+ robot datasets spanning 9 embodiments. In real-world evaluations, PEEK consistently boosts zero-shot generalization, including a 41.4x real-world improvement for a 3D policy trained only in simulation, and 2-3.5x gains for both large VLAs and small manipulation policies. By letting VLMs absorb semantic and visual complexity, PEEK equips manipulation policies with the minimal cues they need--where, what, and how. Website at https://peek-robot.github.io/.
format Preprint
id arxiv_https___arxiv_org_abs_2509_18282
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle PEEK: Guiding and Minimal Image Representations for Zero-Shot Generalization of Robot Manipulation Policies
Zhang, Jesse
Memmel, Marius
Kim, Kevin
Fox, Dieter
Thomason, Jesse
Ramos, Fabio
Bıyık, Erdem
Gupta, Abhishek
Li, Anqi
Robotics
Artificial Intelligence
Machine Learning
Robotic manipulation policies often fail to generalize because they must simultaneously learn where to attend, what actions to take, and how to execute them. We argue that high-level reasoning about where and what can be offloaded to vision-language models (VLMs), leaving policies to specialize in how to act. We present PEEK (Policy-agnostic Extraction of Essential Keypoints), which fine-tunes VLMs to predict a unified point-based intermediate representation: 1. end-effector paths specifying what actions to take, and 2. task-relevant masks indicating where to focus. These annotations are directly overlaid onto robot observations, making the representation policy-agnostic and transferable across architectures. To enable scalable training, we introduce an automatic annotation pipeline, generating labeled data across 20+ robot datasets spanning 9 embodiments. In real-world evaluations, PEEK consistently boosts zero-shot generalization, including a 41.4x real-world improvement for a 3D policy trained only in simulation, and 2-3.5x gains for both large VLAs and small manipulation policies. By letting VLMs absorb semantic and visual complexity, PEEK equips manipulation policies with the minimal cues they need--where, what, and how. Website at https://peek-robot.github.io/.
title PEEK: Guiding and Minimal Image Representations for Zero-Shot Generalization of Robot Manipulation Policies
topic Robotics
Artificial Intelligence
Machine Learning
url https://arxiv.org/abs/2509.18282