Cut apple
OpenPyRo-A1: An Open Python-based Low-Cost Bimanual Robot for Embodied AI
Many real-world tasks, such as assembly, cooking, and object handovers, require bi-manual coordination. However, learning such skills via imitation for these systems remains challenging due to dataset scarcity, driven mainly by the high cost of bi-manual robotic platforms and the barriers to entry in robotics software. To address those challenges, this paper contributes OpenPyRo-A1, a low-cost, bi-manual humanoid robot with a Python-first modular softwarse framework for control, planning, and skill learning. Our system supports VR-based data collection, imitation learning from vision and low-level positions, and integration with LLMs and VLMs for high-level task planning. We evaluate OpenPyRo-A1 on seven bi-manual tasks, collecting over 350 demonstrations via VR teleoperation and showcasing an agentic framework for executing tasks from natural language instructions. We hope that the contributions of the OpenPyRo-A1 hardware, the publicly available software stack, and the curated dataset of bi-manual manipulation episodes will advance affordable, scalable dual-arm robotics.
With our designed distributed robot control framework, IK control algorithm, we can smoothly control the robot to complete the dual-arm cooperative task.
Cut apple
Unpack and fetch
Fold clothes
Wash dishes
Pass the fruit
Open the bottle
Our robot can easily complete long-horizon tasks.
Arrange tableware
Clean board and write
With our high frequency motor control and calibration, our robot can easily complete precise tasks
Write with marker pen
We integrated Action Chunking with Transformers (ACT) and Dynamic Movement Primitives (DMP).
Push apple to the middle of table
Pick and place fruits with two hands
Open cupboard and reach to the target
We present an agent framework that uses a large language model (Deepseek) and a vision-language model (InternVL) to orchestrate a library of learned policies, enabling task execution from natural spoken language instructions. We demonstrate that the system can parse the scene, select relevant skills, and successfully complete a high-level task (e.g., filling a basket with fruit) while ignoring unnecessary actions.
Choose fruits that are yellow in color
Fold clothes
Pour water
We consider the following three key principles when designing OpenPyRo-A1: Low-cost, Ease of repair and Scalability
Low cost printed body
Easy to repair and replace
@misc{huang2025openpyro,
title = {OpenPyRo-A1: An Open Python-based Low-Cost Bimanual Robot for Embodied AI},
author = {Huang, Helong and Mower, Christopher E. and Huang, Guowei
and Das, Sarthak and Dierking, Magnus and Luo, Guangyuan
and Tan, Kai and Chen, Xi and Yang, Yehai and Chen, Yingbing
and Zeng, Yiming and Li, Yinchuan and Zhang, Zhanpeng
and Wu, Shuang and Zhang, Yingxue and Qiu, Weichao
and Cao, Tongtong and Zhuang, Yuzheng and Tian, Guangjian
and Hao, Jianye and Wang, Jun and Bou-Ammar, Haitham
and Quan, Xingyue},
year = {2025},
howpublished = {\url{https://openpyro-a1.github.io/}},
note = {Technical report}
}