SOARM100 is an open-source robotics project. The robot's design is available at The Robot Studio. For training, we leverage algorithms implemented in Hugging Face's LeRobot.
This repository serves as a hub for NONHUMAN projects, including variations, datasets, and experimental developments.
Agentic SOARM100 aims to integrate AI agents, Text-to-Speech (TTS), and Speech-to-Text (STT) models. Once trained with multiple skills, SOARM100 can be controlled intelligently through a preferred Large Language Model (LLM). The interaction is designed to be natural, utilizing a custom interface we have designed instead of relying solely on a command-line interface (CLI), ensuring a more seamless and user-friendly experience.
Neuro SOARM100 explores EEG and EMG signal processing for neural teleoperation. The user’s arm serves as the leading controller, enabling intuitive and direct manipulation of the robotic arm.
For further details, visit our documentation at NONHUMAN Research Wiki.
To set up SOARM100, follow the installation guide available in LeRobot's documentation.
Tip: It is recommended to use
poetryfor dependency management and project setup. Build the project using thepyproject.tomlfile for a streamlined installation.
Detailed guides for both projects can be found in the /guides directory.