POLICY + CERTIFICATION TEAM
01
Every day, roughly 45,000 flights criss-cross above the United States, supported by the diligent work of over 14,000 air traffic controllers.
While staggering, these numbers are just the beginning of what lies ahead: recent estimates project a 4.3% annual increase in air transport demand over the next two decades. As novel advanced air mobility vehicles deploy and scale, our existing Air Traffic Control (ATC) system must adapt to properly serve the increasing complexity of the national airspace.
The bulk of communications between controllers and pilots today is carried out via voice commands, which are essential for ensuring flight safety and facilitating deconfliction in the national airspace and on the airport surface. Yet, conventional voice conversations over radio can lead to misunderstandings, and radio channel congestion can create noisy or blocked transmissions. Even with clear airwaves, pilots sometimes incorrectly write down, misinterpret, or forget lengthy voice communications. And despite increasing complexity and traffic in the airspace, voice commands must be manually registered by controllers, creating increased workload, fatigue, and opportunity for human error.
Artificial intelligence (AI) has the potential to improve system safety and operational performance of the National Airspace System using natural language processing (NLP) to read and interpret human language, and machine learning (ML) to respond. NLP is simply the ability for a computer to understand human language as it’s spoken and written. With NLP, machines can make sense of written or spoken text and perform tasks including speech recognition, sentiment analysis, transcription, and summarization. The automatic transcription of ATC commands using NLP in particular has the immediate potential to improve system safety and operational performance across the aviation landscape by improving efficiency and reducing the opportunity for human error.
At Merlin, we believe AI can play an integral role in aviation, but that a safety framework that balances bleeding-edge ML and traditional software is critical. We are leveraging NLP and ML to provide pilots with additional layers of situational awareness and safety, while also bolstering existing ATC communications. Merlin aims to design a system that will be capable of interacting with ATC over the radio using NLP software. This approach enables real-time analysis and interpretation, which may improve decision-making capabilities for both air traffic controllers and pilots. Automatic speech recognition systems to date have not yet demonstrated the levels of accuracy needed for practical deployment; however, recent developments in ML have led to more accurate speech recognition algorithms.
The adoption of an NLP-powered approach to ATC communications can help us build a more resilient, responsive, and sustainable air traffic management infrastructure by addressing the current challenges addressed earlier in this post. This infrastructure will adeptly accommodate the diverse needs of advanced air mobility vehicles and traditional aircraft, ensuring the safety and efficiency of our skies for generations to come.