10 September 2024 - 11 September 2024, Hall 7, MEETT, Toulouse, France

Leveraging Generative AI for Avionics Software Engineering - Opportunities and Challenges

10 September 2024 14:20 - 14:40

This presentation introduces a visionary approach, leveraging Generative Artificial Intelligence (AI) and Large Language Models (LLMs) to redefine avionics software development

The avionics industry faces a critical challenge: the growing size and complexity of software demands a departure from traditional, manual-intensive engineering methods, which are slow, error-prone, and inefficient. This presentation introduces a visionary approach, leveraging Generative Artificial Intelligence (AI) and Large Language Models (LLMs) to redefine avionics software development.

Generative AI, especially when augmented with LLMs, excels in processing and generating human-like text. Its remarkable adaptability, powered by contextual learning and the potential for domain-specific tuning, offers a strategic advantage in producing accurate and high-quality software lifecycle artifacts—requirements, code, and tests. Moreover, it plays a crucial role in analyzing software defects and field issues, thereby enabling targeted preventive actions that enhance the robustness and reliability of avionics systems.
This capability not only streamlines development but also significantly improves software quality. However, the adoption of Generative AI technology also introduces significant challenges, such as non-determinism, hallucinations, and regulatory and certification hurdles, which require careful consideration. This talk will explore the transformative impact of Generative AI on avionics software engineering, highlighting both the vast opportunities and the challenges to adoption.

Participants will gain a comprehensive understanding of how Generative AI can revolutionize software development processes, ensuring the production of superior avionics software in a more efficient and error-minimized manner.

Abhishek Alladi, Fellow, Verification and Validation CoE, Honeywell Aerospace Technologies