🤓 Latest Submissions
AeroGuard: Autonomous Airfield Voice Interceptor
Commercial airport ramp operations and pushback tug drivers navigate deafening 105+ dB noise environments while maneuvering 300-ton widebody aircraft. They have zero screen availability—looking down at an iPad or tablet while towing aircraft on active taxiways is fatal. Their only lifeline is rapid-fire voice radio. When instructions are misheard, catastrophic collisions occur. In January 2024 at Tokyo Haneda (RJTT), a Coast Guard Dash-8 collided with a landing Japan Airlines A350 at the Taxiway C5 hold line, claiming 5 lives due to a misheard voice clearance. AeroGuard is an autonomous, real-time airfield ground collision and runway incursion interceptor powered by AssemblyAI. AeroGuard continuously monitors ground control radio frequencies, transcribing rapid ICAO clearances with sub-second latency even under extreme turbofan acoustic saturation using AssemblyAI's Streaming Speech-to-Text v3 with domain-specific keyterms prompting. Key Architectural Highlights: 1. Real-Time Clearance & Airfield Grid Engine: Spoken clearances are parsed against an active 2D vector airport grid (Chicago O'Hare and Tokyo Haneda), tracking stopbars, runway safety areas, and conflicting traffic on final approach. 2. Sub-Second Acoustic Barge-In: The moment an unauthorized crossing or hold-line breach occurs, AeroGuard immediately flushes audio buffers and executes an urgent audio alarm and voice directive ("PULL UP! HOLD POSITION!") directly into the operator's headset. 3. ICAO Readback Deviation Engine: Intercepts verbal contradictions before vehicles even move (e.g., ATC commands "Hold short", driver says "Crossing"). 4. Automated FAA Part 139 Digital Audit: Transmits full transcripts and telemetry to the AssemblyAI LLM Gateway to generate structured regulatory compliance logs for safety investigators.
30 Sep 2026
