PURPOSES : This study develops a model that can estimate travel speed of each movement flow using deep-learning-based probe vehicles at urban intersections. METHODS : Current technologies cannot determine average travel speeds for all vehicles passing through a specific real-world area under obseravation. A virtual simulation environment was established to collect information on all vehicles. A model estimate turning speeds was developed by deep learning using probe vehicles sampled during information processing time. The speed estimation model was divided into straight and left-turn models, developed as fully-offset, non-offset, and integrated models. RESULTS : For fully-offset models, speed estimation for both straight and left-turn models achieved MAPE within 10%. For non-offset models, straight models using data drawn from four or more probe vehicles achieved a MAPE of less than 15%. The MAPE for left turns was approximately 20%. CONCLUSIONS : Using probe-vehicle data(PVD), a deep learning model was developed to estimate speeds each movement flow. This, confirmed the viability of real-time signal control information processing using a small number of probe vehicles.

PURPOSES : The purpose of this study is to build an optimization model using the capacity and initial travel speed of the volume delay functions for network calibration performed in the traffic demand analysis process. METHODS : The optimization model contains an error term between the observed traffic volume and estimated traffic volume, based on the user equilibrium principle, and was constructed as a bi-level model by applying range constraints on capacity and travel time. In addition, we searched the split section to apply the method of adjusting the section instead of adjusting the single link. The optimization model is constructed by applying the warm-start method using the bush of the origin-based model so that parameter adjustment and traffic assignment are repeatedly executed within the model and the convergence of the model configured %RSSE. RESULTS : As a result of analysis using the toy network, the optimization model is that the observed traffic volume is estimated when there are no restrictions and, when the constraint conditions were set, the error with the observed traffic volume and error rate was significantly reduced. As a result of the comparative analysis of the trial-and-error methods, KTDB optimum values, and optimization models in empirical analysis using a large-scale network, the evaluation indexes (e.g., RMSE and %RMSE) were significantly improved by applying the optimization model. CONCLUSIONS : Based on the empirical analysis, the optimization model of this study can be applied to large-scale networks and it is expected that the efficiency and reliability of road network calibration will be improved by repeatedly performing parameter adjustment and traffic assignment within the model.

Recently, Live-Virtual-Constructive (L-V-C) integrate training system has proposed as a solution for the problems such as limitation of training areas, increase of mission complexity, rise in oil prices. In order to integrate each training system into the one effectively, we should solve the issue about stress of pilots by the environmental differences between Live and Virtual simulation which could be occurred when each system is connected together. Although it was already examined in previous study that the psychological effects on pilots was occurred by the environmental differences between actual and simulated flights, the study did not include what the causal factors affecting psychological effects are. The aim of this study is to examine which environmental factors that cause pilots’ psychological effects. This study analyzed the biochemical stress hormone, cortisol to measure the pilots’ psychological effects and cortisol was measured using Enzyme-linked immunoassay (EIA). A total of 40 pilots participated in the experiment to compare the differences in pilots’ cortisol response among live simulation, virtual simulation, and the virtual simulation applying three environmental factors (gravity force, noise, and equipment) respectively. As a result, there were significant differences in cortisol level when applied the gravity force and equipment factors to the virtual simulation, while there was no significant difference in the case of the noise factor. The results from this study can be used as a basis for the future research on how to make L-V system by providing minimum linkage errors and design the virtual simulator that can reduce the differences in the pilots’ psychological effects.

Recently, Korea Air Force has been facing a lot of problems in its pilot training system such as training time shortage due to the expensive gas price, noise pollution and difficulties in finding airspace for training. To tackle these problems, a new training system (called L-V training system) using both aircraft and its simulator has been suggested. In the system, a data link is established between aircraft and simulator to exchange their flight information. Using the flight information of simulator, aircraft can perform various air missions with or against imaginary aircraft (i.e., simulator). For this system, it is crucially important that fair fighting condition has to be guaranteed between aircraft and simulator. In this paper, we suggested an approach to impose a maneuvering restriction to simulator in order to provide fair fighting condition between aircraft and simulator.