Abstract
There are clear opportunities in exploiting artificial intelligence (AI) in air traffic control. One of the relevant areas where AI has a potential is in supporting air traffic controllers in their work, particularly in periods with high workload. A possible approach to providing such support is to have a digital assistant as part of the air traffic controllers (ATCOs) support tools. To investigate how such an assistant may be designed, we have developed the DIALOG system. This system supports ATCO when responding to pilot requests. An Intent Inferring System listens to the radio communication between ATCO and pilot, and based on pilot utterances, it identifies the request and collects the most important information needed by the ATCO to respond to the request. A Teammate Awareness Service measures physiological data from ATCOs and uses machine learning models to induce the ATCOs workload in an unobtrusive manner. A Teamwork Assistant uses information from the Teammate Awareness Service to determine which type of assistance to provide to the ATCO at any given time. When a pilot issues a request, the Teamwork Assistant presents information from the Intent Inferring System in a manner suiting the ATCO’s current state. In this paper we present the different parts of the DIALOG system, focusing on the interplay between the Teamwork Assistant and the ATCO in the user interface of the ATCO. We also explain how the assistant was validated and discuss the main challenges and experiences faced while designing its user interface.