What's Happening?
NASA's Ames Research Center is advancing a project to modernize cockpit communications by integrating edge AI and commercial large language models (LLMs). The initiative, part of the Flight Deck Collaborative Decision Making (CDM) Research Project, aims
to transition from analog, voice-based radio commands to a digital data environment. The project involves digitizing Very High Frequency (VHF) aircraft taxi commands and converting them into digital instructions using a Google Speech-to-Text LLM. These instructions will be streamed to pilots' electronic flight bags and air traffic control networks. The project is in its fifth phase, focusing on real-time edge computing, utilizing a Raspberry Pi processing core and custom audio electronics to minimize latency in airport operations.
Why It's Important?
This development is significant as it represents a shift towards more efficient and reliable communication systems in aviation, potentially reducing errors associated with voice-based instructions. By digitizing communications, NASA aims to enhance the safety and efficiency of aircraft surface movements. The integration of edge AI and LLMs could set a precedent for future advancements in aviation technology, influencing how airlines and air traffic control manage operations. This could lead to broader adoption of digital communication systems across the aviation industry, improving operational efficiency and safety.
What's Next?
NASA plans to conduct shadow and live testing of the new system, with a 'Shadow Mode' test scheduled for November 2026 and a live-assessment test in March 2027. These tests will evaluate the system's performance in real-time scenarios, focusing on latency and word-error-rate metrics. The outcomes of these tests will determine the feasibility of implementing the system on a larger scale. NASA is also open to receiving input from other organizations interested in contributing to the project's development.
