This repository offers a curated set of prompts for leveraging ChatGPT in robot manipulation tasks, enabling natural language instruction interpretation and translation into executable robot actions. It targets researchers and developers seeking to integrate large language models with robotic systems for enhanced human-robot interaction.

How It Works

The core approach involves crafting specific prompts that guide ChatGPT to decompose complex, long-step instructions into a sequence of actionable robot commands. This method facilitates seamless integration with existing robot control frameworks and visual recognition systems, abstracting away low-level control details.

Quick Start & Requirements

• Install dependencies: pip install -r requirements.txt
• Run sample code: python aimodel.py --scenarios <scenario_name> from within an examples/ subfolder.
• Prerequisites: Python 3.9.16 confirmed working. Requires OpenAI or Azure OpenAI API credentials.
• Documentation: Blog post and research papers are linked in the README.

Highlighted Details

• Supports OpenAI and Azure OpenAI APIs.
• Sample code provided for task decomposition scenarios.
• Prompts are designed for customization and integration.
• Research papers detailing the methodology are available.

Maintenance & Community

This project is from Microsoft and welcomes contributions via pull requests, subject to a Contributor License Agreement (CLA). It adheres to the Microsoft Open Source Code of Conduct.

Licensing & Compatibility

The repository does not explicitly state a license. However, given its origin from Microsoft and the use of OpenAI APIs, users should verify licensing implications for commercial use or integration into closed-source systems.

Limitations & Caveats

The effectiveness of the prompts is dependent on the capabilities of the underlying LLM and the specific robot's action space. The sample code requires configuration of API secrets and scenario selection.