Proactive Improvements of Standard Taxi Routes in Airports Based on Data-Driven Simulation and Reinforcement Learning
To relieve the workload of Air Traffic Controllers (ATCs) and reduce radio frequency congestion, many busy airports adopt Standard Taxi Routes (STR) in their daily operations. STRs are predefined taxi routes published to ATCs and flight crews. However, due to the uncertainty and dynamic changes in the operations, the lack of guidelines for adjusting STRs in different traffic scenarios makes it hard for ATCs to make effective adjustments. Automatically adjusting safe and resilient taxi routes with uncertain and changing traffic is challenging. This research established a data-driven simulation augmented by reinforcement learning algorithms to support the taxiing route design work. The proposed research has three contributions: 1) a fast and reliable map matching method for reconstructing daily operation; 2) a data-driven simulation platform for airport traffic simulations; 3) characterization of the control strategies for multi-objective STRs planning and adjustments through different settings of reinforcement learning (RL). This research used the Los Angeles International Airport as an example to build the map matching, simulation, and optimization of airport ground operation. The results show that 1) the proposed map matching method achieved 100% accuracy for one hundred aircraft trajectories in the LAX airport; 2) the developed simulation platform can predict conflict hot spots, which covers 100% of the hot spots identified from historical records; 3) the established RL approach has been tested in two traffic scenarios, i.e., normal traffic and heavy traffic with conflict hot spots, the established RL approach can effectively adjust the taxiing route for aircraft under different scenarios.
- Record URL:
-
Supplemental Notes:
- This paper was sponsored by TRB committee AV060 Standing Committee on Airfield and Airspace Performance. Alternate title: ACRP Graduate Research Award: Proactive Improvements of Standard Taxi Routes in Airports Based on Data-Driven Simulation and Reinforcement Learning
-
Corporate Authors:
Transportation Research Board
, -
Authors:
- Wang, Yanyu
- Tang, Pingbo
-
Conference:
- Transportation Research Board 102nd Annual Meeting
- Location: Washington DC, United States
- Date: 2023-1-8 to 2023-1-12
- Date: 2023
Language
- English
Media Info
- Media Type: Digital/other
- Features: Figures; Maps; References; Tables;
- Pagination: 17p
Subject/Index Terms
- TRT Terms: Air traffic control; Ground handling; Machine learning; Routes; Simulation; Taxiing
- Identifier Terms: Los Angeles International Airport
- Geographic Terms: Los Angeles (California)
- Subject Areas: Aviation; Data and Information Technology; Operations and Traffic Management; Safety and Human Factors;
Filing Info
- Accession Number: 01875179
- Record Type: Publication
- Report/Paper Numbers: TRBAM-23-01808
- Files: TRIS, TRB, ATRI
- Created Date: Mar 1 2023 10:09AM