Airport Taxi-Out Prediction Using Approximate Dynamic Programming

Abstract

Flight delay is one of the pressing problems that have far-reaching effects on society and the nation's economy. A primary cause of flight delay in the National Airspace System is high taxi-out times (time between gate push-back and wheels-off) at major airports. Accurate prediction of taxi-out time is needed to make downstream schedule adjustments and for better departure planning, which could mitigate delays, emissions, and congestion on the ground. However, accurate prediction of taxi-out time is difficult because of uncertainties associated with the dynamically changing airport operation. A novel stochastic approximation scheme based on reinforcement learning (RL) is presented for predicting taxi-out times in the presence of weather and other departure-related uncertainties. The prediction problem is cast in a probabilistic framework of stochastic dynamic programming and solved by using approximate dynamic programming approaches (particularly RL). The strengths of the method are that it is nonparametric, unlike regression models with fixed parameters, it is highly adaptable to the dynamic airport environment since it is learning based, it is scalable, it is inexpensive since it does not need highly sophisticated surface management systems, and it can effectively handle uncertainties because of the probabilistic framework. Taxi-out prediction performance was tested on data obtained from the FAA Aviation System Performance Metrics database on Detroit International and Washington Reagan National Airports. Results show that the root-mean-square prediction error calculated 15 min before gate departure time is on average 2.9 min for about 80% of the predicted flights.

Landing on empty: estimating the benefits from reducing fuel uplift in US Civil Aviation

“…Regarding ground-based initiatives, consider that there are numerous studies and federal initiatives focused on reducing fuel consumed by an aircraft during taxi out (from the gate to the runway) (Daniel 2002, Balakrishna et al 2008, Nikoleris et al 2011, Khadilkar and Balakrishnan 2012, Hao et al 2015. Simaiakis et al (2014) find that managing the rate of aircraft pushback from the gate a busy airport could reduce per-flight fuel consumption by about 110-130 lbs of fuel, a value slightly less than the median value of fuel savings from reducing ACAF from a flight.…”

Airport taxi planning: Lagrangian decomposition

“…The tool is designed to adapt to perturbations in these input conditions and account for failure in the actual execution of surface trajectories. [14][15][16], have investigated the taxi-out prediction by using a nonparametric reinforcement learning-based method set in the probabilistic framework of stochastic dynamic programming.…”

“…Some authors, such as Balakrishna et al , have investigated the taxi‐out prediction by using a nonparametric reinforcement learning‐based method set in the probabilistic framework of stochastic dynamic programming.…”

Improving quality of prediction in highly dynamic environments using approximate dynamic programming

“…In this research paper, the prediction problem is modeled as an MDP, and solved using RL principles. In the prior work, the Q‐learning (Gosavi26) approach was used for predicting taxi‐out times at Detroit and Tampa Bay airports using look‐up tables that stored the Q‐factors35, 36. This paper presents in detail the ADP learning algorithm using V ‐values, which alleviates the computational burden of storing Q‐factors, and tests the algorithm for predicting taxi‐out times at several airports.…”