Sehyun Tak

Development of a Deceleration-Based Surrogate Safety Measure for Rear-End Collision Risk

A surrogate safety measure can be used for preventing hazardous roadway events by evaluating the potential safety risk by using information on the driving environment gathered from vehicles. In this paper, the deceleration-based surrogate safety measure (DSSM) is proposed as a safety indicator for rear-end collision risk evaluation based on the safety conditions and the decision-making process during human driving. The DSSM shows how drivers deal with collision risk differently in acceleration and deceleration phases. The proposed surrogate safety model has been validated for severe deceleration behavior, which is a driver-critical behavior in high-risk situations of collision based on microscopic vehicle trajectory data. The results indicate that there is a strong relationship between the proposed surrogate safety measures and crash potential. The measure could be used for collision warning and collision avoidance systems. It has a merit in that it reflects the characteristics of both vehicle (e.g., mechanical braking capability) and driver (e.g., preference for certain acceleration rates).

Real-Time Travel Time Prediction Using Multi-Level k-Nearest Neighbor Algorithm and Data Fusion Method

Estimating and predicting the travel time on freeways with reasonable accuracy is essential for successful implementation of intelligent transportation system. However, the related previous studies have some problems. The statisticsbased methods have problems in accuracy, and some others are limited to predicting the travel time only during short time interval. Another challenging matter is that the existing road sensors have some limitations in being directly utilized, because data from road sensors have lots of errors. In this study we propose a new algorithm called multi-level k-Nearest Neighbor (k-NN), which is designed for predicting travel time with higher computational efficiency and prediction accuracy. The algorithm consists of three parts: classification, global matching, and local matching. As a part of the proposed algorithm, in order to overcome the problems of data errors, we provide a data fusion method that combines the traffic data from ILDs and DSRC. The results show that the proposed multi-level k-NN with the data fusion can effectively predict the future travel time within less than 5% error range, even in congested traffic

Data-Driven Imputation Method for Traffic Data in Sectional Units of Road Links

Missing data imputation is a critical step in data processing for intelligent transportation systems. This paper proposes a data-driven imputation method for sections of road based on their spatial and temporal correlation using a modified k- nearest neighbor method. This computing-distributable imputation method is different from the conventional algorithms in the fact that it attempts to impute missing data of a section with multiple sensors that have correlation to each other, at once. This increases computational efficiency greatly compared with other methods, whose imputation subject is individual sensors. In addition, the geometrical property of each section is conserved; in other words, the continuation of traffic properties that each sensor captures is conserved, therefore increasing accuracy of imputation. This paper shows results and analysis of comparison of the proposed method to others such as nearest historical data and expectation maximization by varying missing data type, missing ratio, traffic state, and day type. The results show that the proposed algorithm achieves better performance in almost all of the missing types, missing ratios, day types, and traffic states. When the missing data type cannot be identified or various missing types are mixed, the proposed algorithm shows accurate and stable imputation performance.

A Study of the Framework on Collision Risk Warning System Using Loop Detector and Vehicle Information

Current safety warning systems generally operate based on the information from sensors attached to individual vehicles. This vehicle sensor-based system can only estimate the collision potential situation in close proximity of a subject vehicle, and it requires additional communication technologies such as Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication technologies in order to obtain a wide range of information. The device requirements for such technologies will lead to the price increase of the collision warning system. So, in this study, the authors propose the collision warning system that utilizes the information from the fusion of loop detector data and smartphone data. The proposed collision warning system can be directly applied without any additional cost because the databases of loop detector and smartphones are available to be used. The developed system is tested by simulating a real vehicle trip based on the NGSIM data and comparing its results to vehicle-sensor-based system and infrastructure information-based system. It was found that the newly proposed collision risk warning system can show the similar performance with the V2V-based collision warning system.

The Impact of Predictive Cruise Control on Traffic Flow and Energy Consumption

A vehicle Predictive Cruise Control system has been developed to improve the fuel efficiency of vehicle and traffic flow performance based on the asymmetric traffic theory. The Predictive Cruise Control system consists of four parts: (1) Deceleration based Safety Surrogate Measure, (2) Adaptive Cruise Control, (3) Multi-vehicle measurement, and (4) Predictive Cruise Control. Adaptive Cruise Control basically decides the acceleration/deceleration action based on the estimated deceleration-based safety surrogate measure of the first leader vehicle. Then, Predictive Cruise Control adjusts the acceleration/deceleration action based on the multi-vehicle measurements, which represent the future traffic condition of the subject vehicle. The developed system is tested by simulating the real vehicle trajectories from the NGSIM data and comparing the results with real following pattern. It was found that the newly proposed Predictive Cruise Control system can contribute to energy consumption and traffic flow performance, because it can effectively suppress the shockwave from the downstream and remove the unnecessary deceleration and acceleration action.

Data-Driven Prediction Methodology of Origin-Destination Demand in Large Network for Real-Time Service

Prediction of origin–destination (O-D) demand is an important topic in transportation engineering because it is a crucial input for a dynamic traffic management and control system. Previous literature has focused primarily on estimation and prediction of O-D demand with Kalman filtering; however, these forecasts lack efficiency when unusually fluctuating O-D demand of a large O-D network is predicted in real time. With true, real-time O-D demand data from South Korean expressways, a data-driven prediction framework of O-D demand in a large network for real-time service is proposed by modifying the k–nearest neighbor (k-NN) algorithm. Three strategies that implement different feature vectors for k-NN prediction of single-level O-D demand, multilevel O-D demand, and single-level point demand are proposed. The strategies were tested on hourly O-D demand in South Korea. The average mean absolute percentage of error values of the three strategies in terms of total demand are 5.52%, 5.34%, and 3.36%, respectively; single-level point demand performs slightly better than do the other two strategies. Similarly, for the average mean absolute percentage and weighted average mean absolute percentage in terms of individual O-D demand, single-level point demand performs better than do the two other strategies, especially for O-D pairs with larger demand and for further prediction horizons. In addition, the single-level point demand shows the highest computation efficiency. Therefore, the single-level point demand strategy for k-NN prediction shows the best combination of accuracy and computation efficiency among the three strategies. Furthermore, a historical database size of at least 300 dates for this data-driven prediction algorithm seems required for accuracy.

Development of a Data-Driven Framework for Real-Time Travel Time Prediction

Travel time prediction is one of the most important components in Intelligent Transportation Systems implementation. Various related techniques have been developed, but the efforts for improving the applicability of long-term prediction in a real-time manner have been lacking. Existing methods do not fully utilize the advantages of the state-of-the-art cloud system and large amount of data due to computation issues. We propose a new prediction framework for real-time travel time services in the cloud system. A distinctive feature is that the prediction is done with the entire data of a road section to stably and accurately produce the long-term (at least 6-hour prediction horizon) predicted value. Another distinctive feature is that the framework uses a hierarchical pattern matching called Multilevel k-nearest neighbor (Mk-NN) method which is compared with the conventional k-NN method and Nearest Historical average method. The results show that the method can more accurately and robustly predict the long-term travel time with shorter computation time.

Large-Scale Maintenance Optimization Problems for Civil Infrastructure Systems

Maintenance optimization based on life cycle cost assessment is applied to many civil infrastructure systems such as pavements and bridges to reduce operating and socioeconomic costs. The importance of efficient and economical solutions for maintaining large-scale systems is increasing with the soaring need for the sustainable infrastructure systems from climate change and budget constraints from the worldwide economic crisis. This chapter addresses the state-of-the-art algorithms for multi-facility large-scale infrastructure maintenance problem issues and solutions. Then it provides an example of case study for a large-scale problem applying metaheuristics algorithms such as genetic algorithm, particle swarm optimization, ant colony optimization, shuffled frog leaping, pattern search heuristic, and evolutionary algorithm.

DRIVERS’ EYE GLANCE TRANSITIONS AND THE IMPLICATIONS ON THE MICROSCOPIC TRAFFIC AND ACCIDENT SIMULATION

Driver’s inattention and distraction are one of the main causes of traffic accidents. Therefore, understanding on the driver’s behavior of inattention and distraction gives important implications for improving traffic safety. Previous researches tried to identify the main factors leading to accidents and their impact on traffic accidents. However, it is hard to model and estimate the impact as the drivers’ distraction shows stochastic transitions. In this study, we try to provide a better understanding on the relationship between distraction and vehicle safety and analyze the dynamics of driver vision transitions in both normal driving situation and accident situation. Based on the result, we propose a new car-following model, which can reflect the driver’s dynamic vision transition, which can regenerate the risk situations that the existing car-following models cannot provide. The new model can provide a framework for better understanding on the relationship of traffic accidents and drivers vision transition.

Agent-based pedestrian cell transmission model for evacuation

ABSTRACT Describing dynamics of pedestrian evacuation is a challenging issue. Many related studies have been conducted in both microscopic and macroscopic views. Microscopic approaches focus on detailed movements of individual pedestrians and they are computationally expensive. Macroscopic approaches have less concern in the individuals, but they are lack of flexibility because they treat pedestrians as unthinking elements that cannot choose their destinations or routes. We propose the agent-based pedestrian cell transmission model (A-PCTM) that puts the merits of both microscopic and macroscopic approaches. We modify the original CTM to improve the computational efficiency and embed the concept of agents that make decisions on their destinations and travel directions. The A-PCTM has procedures of adaptive direction-based path finding, and the effect of such procedures is tested by some simulation cases with different geometric settings. The proposed model shows the flexibility in switching the destination and in choosing the travel direction depending on the situation ahead.