24 August 2021 in Transportation Research Record: Journal of the Transportation Research Board
A significant fraction of communications between air traffic controllers and pilots is through speech, via radio channels. Automatic transcription of air traffic control (ATC) communications has the potential to improve system safety, operational performance, and conformance monitoring, and to enhance air traffic controller training. We present an automatic speech recognition model tailored to the ATC domain that can transcribe ATC voice to text. The transcribed text is used to extract operational information such as call-sign and runway number. The models are based on recent improvements in machine learning techniques for speech recognition and natural language processing. We evaluate the performance of the model on diverse datasets.
1 July 2020 in Transportation Science
Congestion at major airports worldwide results in increased taxi times, fuel burn, and emissions. Regulating the pushback of aircraft from their gates, also known as departure metering, is a promising approach to mitigating surface congestion. Departure metering algorithms require models of airport surface traffic and knowledge of when a flight would be to be ready for pushback, which is called the earliest off-block time (EOBT). While EOBTs are known to be inaccurate due to several reasons, there has been little prior research on characterizing EOBT uncertainty and its impact on departure metering. We present a new class of queuing network models for the airport surface that are capable of capturing congestion at multiple locations. We demonstrate our modeling approach using operational data from three major U.S. airports: Newark Liberty International Airport, Dallas/Fort Worth International Airport, and Charlotte Douglas International Airport. We analyze the current levels of uncertainty in the EOBT information published by the airlines and conduct a parametric analysis of the reduction in departure metering benefits due to errors in the EOBT information. Our analysis indicates that the current levels of EOBT uncertainty lead to a 50% reduction in benefits at some airports when compared with an ideal case with no EOBT uncertainty. Two approaches to departure metering are considered: the National Aeronautics and Space Administration’s Airspace Technology Demonstration-2 logic and a new optimal control approach. We show that our queuing network models can help design and evaluate both approaches and that the optimal control approach is more effective in accommodating EOBT uncertainty while maintaining runway utilization.
1 January 2021 in IEEE Transactions on Intelligent Transportation Systems
Queuing networks have been widely-used to model congestion in transportation systems. Due to their interconnected nature, delays in a queuing network can propagate as customers traverse through the network; similarly, downstream resources can be underutilized due to poor control policies. This paper considers the regulation of arrivals into a queuing network in order to maintain a desired level of occupancy (queue length) in the system. The dynamics of the queuing network is represented by a fluid-flow model, which is then used to develop a robust controller for tracking the desired queue length. The controller is based on a sliding mode control approach, with predictor-based feedback to account for propagation delays. For a single queue, we determine sufficient conditions for tracking the queue length, and bounds on the tracking error. We also present an analysis of the tracking performance for queues in tandem. We demonstrate our approach for the example of airport surface congestion control. The proposed robust control framework is based on a queuing network model of the airport, and is used to tactically manage aircraft departures in order to reduce congestion on the airport tarmac.
6 March 2021 in CEAS Aeronautical Journal
1 July 2023 in Journal of Air Transportation
The assessment of the fuel burn and emissions impact of airport surface operations is a key part of understanding the environmental impacts of aviation. These assessments are needed at two levels: the analysis of inventories (the total amount of fuel burned and emissions discharged over some period of time), and the analysis of spatial distributions (the amount of emissions experienced at a particular location within or near the airport). Although the availability of taxi times for the operations of interest is sufficient for inventory analysis, the analysis of spatial distributions requires estimates of where on the airport surface an aircraft is located as it consumes fuel. In this paper, we show how a data-driven queuing network model can be developed in order to estimate the time that an aircraft spends at different congested locations on the airport surface. These models are useful both in spatial distribution analysis and in accurately predicting taxi times in the absence of measurements (for example, for projected demand sets). We use measurements of ultrafine particles at Los Angeles International Airport to demonstrate that the proposed model can help predict the measured emissions at different monitoring sites located in the vicinity of the airport. In the process, we show how one could develop a machine learning model of the spatial distribution of airport surface emissions given the pollutant measurements, air traffic demand, and prevailing weather conditions. Finally, we develop a clustering-based method to evaluate the generalizability of our surface operations modeling framework.