cheng-haha/GPT-Policy

This repository is the public implementation of GPT-Policy, a closed-loop control framework that connects a fixed vision-language model (VLM) to robot tools. At deployment time, the agent can use demonstrations, goal images, interaction history, and execution feedback without gradient updates or task-specific parameter changes. The implementation provides hardware adapters for ARX X5 and I2RT/YAM, plus an optional RoboDojo simulation backend with a separate Isaac process. [](https://cheng-haha.github.io/GPT-Policy/) [](https://arxiv.org/abs/2609.19138) [](https://arxiv.org/pdf/2609.19138) [](https://x.com/zcode68632/status/2098397364725895387) [](https://huggingface.co/datasets/SII-CDZ/gpt-policy-robodojo-contexts) Paper: In-Context Robot Learning with VLM Agents \ Authors: Dongzhou Cheng, Taoran Yi, Ye Fang, Xingwu Zhang, Fan Feng, Yixuan Li, Gengxiong Zhuang, Rongze Wang, Shuai Yang, Wei Song, Weizhi Xue, Minyan Wu, Jie Gui, Jiaqi Wang, and Tong Wu. Abstract Four synchronized views of three GPT-6 Astra robot runs. Click any preview to open its MP4. Plug insertion · top view      Plug insertion · right wrist view Sprite retrieval · search and place the bottle      Bottle opening…