Saved in:
Bibliographic Details
Main Authors: Zhao, Ziyang, Wang, Shuheng, Miao, Zhonghua, Xiong, Ya
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.05982
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911492659478528
author Zhao, Ziyang
Wang, Shuheng
Miao, Zhonghua
Xiong, Ya
author_facet Zhao, Ziyang
Wang, Shuheng
Miao, Zhonghua
Xiong, Ya
contents This work presents the first study on transferring vision-language-action (VLA) policies to real greenhouse tabletop strawberry harvesting, a long-horizon, unstructured task challenged by occlusion and specular reflections. We built an end-to-end closed-loop system on the HarvestFlex platform using three-view RGB sensing (two fixed scene views plus a wrist-mounted view) and intentionally avoided depth clouds and explicit geometric calibration. We collected 3.71 h of VR teleoperated demonstrations (227 episodes) and fine-tuned pi_0, pi_0.5, and WALL-OSS with full fine-tuning and LoRA. Under a unified 50 trials real-greenhouse protocol and metrics spanning completion, pi_0.5 with full fine-tuning achieved success rate of 74.0% with 32.6 s/pick and damage rate of 4.1%. Asynchronous inference-control decoupling further improved performance over synchronous deployment. Results showed non-trivial closed-loop picking with fewer than four hours of real data, while remaining limited by close-range observability loss and contact-dynamics mismatch. A demonstration video is available at: https://youtu.be/bN8ZowZKPMI.
format Preprint
id arxiv_https___arxiv_org_abs_2603_05982
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HarvestFlex: Strawberry Harvesting via Vision-Language-Action Policy Adaptation in the Wild
Zhao, Ziyang
Wang, Shuheng
Miao, Zhonghua
Xiong, Ya
Robotics
Computer Vision and Pattern Recognition
This work presents the first study on transferring vision-language-action (VLA) policies to real greenhouse tabletop strawberry harvesting, a long-horizon, unstructured task challenged by occlusion and specular reflections. We built an end-to-end closed-loop system on the HarvestFlex platform using three-view RGB sensing (two fixed scene views plus a wrist-mounted view) and intentionally avoided depth clouds and explicit geometric calibration. We collected 3.71 h of VR teleoperated demonstrations (227 episodes) and fine-tuned pi_0, pi_0.5, and WALL-OSS with full fine-tuning and LoRA. Under a unified 50 trials real-greenhouse protocol and metrics spanning completion, pi_0.5 with full fine-tuning achieved success rate of 74.0% with 32.6 s/pick and damage rate of 4.1%. Asynchronous inference-control decoupling further improved performance over synchronous deployment. Results showed non-trivial closed-loop picking with fewer than four hours of real data, while remaining limited by close-range observability loss and contact-dynamics mismatch. A demonstration video is available at: https://youtu.be/bN8ZowZKPMI.
title HarvestFlex: Strawberry Harvesting via Vision-Language-Action Policy Adaptation in the Wild
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.05982