Enhancing the development of Cherenkov Telescope Array control software with Large Language Models

2025-10-01

AI Summary

This study addresses challenges in the development and operation of the Control and Data Acquisition (ACADA) software for the Cherenkov Telescope Array Observatory (CTAO), including fragmented domain knowledge, complex interfaces, and low cross-team collaboration efficiency. We propose an AI agent framework based on instruction-tuned large language models (LLMs), tightly integrating project documentation, source code repositories, and astronomical instrument APIs to enable context-aware natural language interaction and automated task execution. Our key contribution is the first end-to-end integration of instruction-tuned LLMs across the full software engineering lifecycle of astronomical observation control systems—spanning development assistance, real-time operational monitoring, and offline data analysis. Experimental evaluation demonstrates significant improvements in code comprehension accuracy, documentation retrieval efficiency, and inter-module collaboration latency, thereby enhancing system maintainability and engineering scalability.

Technology Category

• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Multiagent Systems: Agent Communication
• Humans and AI: Planning and Decision Support for Human-Machine Teams

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

We develop AI agents based on instruction-finetuned large language models (LLMs) to assist in the engineering and operation of the Cherenkov Telescope Array Observatory (CTAO) Control and Data Acquisition Software (ACADA). These agents align with project-specific documentation and codebases, understand contextual information, interact with external APIs, and communicate with users in natural language. We present our progress in integrating these features into CTAO pipelines for operations and offline data analysis.

Problem

• Developing AI agents for CTAO control software engineering
• Integrating LLMs to understand documentation and codebases
• Enhancing telescope operations with natural language interaction

Innovation

• AI agents based on fine-tuned large language models
• Align with project documentation and codebases
• Interact with APIs and communicate in natural language