A Speech-Enabled Simulation Interface Agent for Airspace System Assessments

Abstract

Extensive human-in-the-loop testing of the Next-Generation Air Transportation System (NextGen) concepts and technologies is typically required in a controlled lab environment before they can be integrated for evaluation in the field. The experiments tend to require the participation of a large number of subject-matter experts (SMEs) including air traffic controllers (ATC) and (pseudo-)pilots, where such requirements make the experiments costly and the logistics with so many participants make them difficult to plan. These experiments often are designed only to collect data from either ATC or the pilots, but seldom both; the counterpart is needed only to provide realism in communication between them. The paper illustrates our research in developing a Speech-Enabled Simulation Interface Agent (SESIA) to replace the non-essential human subjects in these experiments. SESIA can interact with the SMEs through voice communication, and interface with the simulation platform to perform the intended control. It has the benefit of reduced cost associated with the experiments and increased convenience in their planning, thus allowing the opportunities to schedule additional testing. In cases where a pseudo-pilot would normally represent multiple flights and communicate to the ATC with the same voice for all flights, SESIA will actually increase the realism of the experiments by allowing different voices to be synthesized to simulate different pilots.