ASTRA Simulator Automates Air Traffic Control Training with AI Simpilots

Air Traffic Control Operator (ATCO) training traditionally relies on human "simpilots" who role-play both pilots and other ATCOs, limiting training capacity. Existing automated solutions struggle with diverse accents, particularly in regions like Singapore, where speech recognition accuracy is poor. Researchers have developed ASTRA, a new training simulator designed to automate these simpilot roles. ASTRA employs a sophisticated AI pipeline that transcribes ATCO speech, interprets instructions, and generates appropriate pilot and ATCO responses using voice models specifically adapted for local accents. This system not only enhances traffic simulation but also includes an AI-assisted framework for evaluating trainee radiotelephony communications across accuracy, brevity, and completeness. Built on open-source tools, ASTRA aims to provide scalable, standardized ATCO assessment while reducing the workload on human instructors.