Hyunmyung Kim

Origin-Destination Matrices Estimated with a Genetic Algorithm from Link Traffic Counts

Several approaches have been developed to cope with the limits of conventional origin-destination (O-D) trip matrix collecting methods. One is the bilevel programming method, which uses a sensitivity analysis-based (SAB) algorithm to solve a generalized least-squares problem. However, the SAB algorithm has revealed a critical shortcoming when there is a significant difference between the target O-D matrix and the true O-D matrix. This problem stems from the heavy dependence of the SAB algorithm on historical O-D information. Such dependence may lead to a state in which the O-D estimator cannot produce a correct solution, especially when travel patterns are dramatically changed. To avoid the problem of dependency, a robust and stable method is required. A solution method is developed with a genetic algorithm, which is widely used in optimization problems to obtain a global solution. From the results of numerical examples, the proposed algorithm is superior to the SAB algorithm regardless of travel patterns.

Application of Activity Chaining Model Incorporating a Time Use Problem to Network Demand Analysis

An activity chaining model described as a time use problem is presented. Unlike existing activity chaining models, the time use problem includes all choice attributes related to time, not only when and where the activity is performed but also the duration of the activity. To apply the time use problem to transportation network demand analysis, the opportunity cost of travel time is modeled through a constraint approach, and the value of activity is represented by time-dependent utility profiles. In the model, among activity alternatives, the activity with the biggest utility is sequentially included within an individuals daily disposable income. Through the sequential selection process, not only the activity type but also its start time and duration are simultaneously determined on the basis of an individuals perceived time-dependent travel time for the activity. Since the activity chaining is performed on the basis of the dynamic travel times calculated via traffic network simulation, the activity chai...

Dynamic Origin-Destination Estimation Using Dynamic Traffic Simulation Model in an Urban Arterial Corridor

As dynamic traffic simulation (DTS) becomes ever more popular, how to provide the input data, that is, a dynamic origin–destination (O-D) table, for simulation becomes an important question. During recent years, many researchers have proposed estimation methods for a dynamic O-D table from an academic point of view, but transportation engineers appear to use the dynamic approach in practical applications rarely. In recent simulation projects, ad hoc schemes have been used by practitioners for developing dynamic O-D tables for subareas from larger, static planning O-D tables without much theoretical soundness. The authors developed a dynamic O-D estimation model based on a path-flow approach and iterative simulations, along with a scheme to find a justifiable seed O-D table, and applied it to a real-world context. A major arterial corridor in Suwon, South Korea, was selected, and the link turning counts were collected at intersections. A microscopic DTS model, PARAMICS, was used with the dynamic O-D table to simulate link turning counts. The test results show that the developed O-D estimation technique is efficient, with simulated link traffic counts and section travel times very close to the reality.

Analysis of Grid Cell–Based Taxi Ridership with Large-Scale GPS Data

Understanding the spatial variation of taxi ridership is of critical importance to many government agencies and taxi companies because taxis’ location dependency on spatial pattern of passenger demand results in spatially unbalanced taxi demand and supply. This study presents an analysis of the spatial distribution of taxi ridership by using large-scale GPS taxi trip data collected from Seoul, South Korea. To capture the spatial variations better in taxi ridership, GPS entities were disaggregated into units of a uniform size with a grid cell decomposition method. A geographically weighted spatial regression was applied to model spatial correlations of factors associated with transit and urban density to taxi ridership. Results from the proposed method demonstrated a higher relationship between taxi and subway ridership in the regions where lower accessibility to subway stations existed. In these regions, taxis were found to perform as a complementary mode to subway. In residential and commercial districts, this analysis showed that population and employment were highly related to taxi ridership. In contrast, in central business districts it was the building area (floor space), rather than population and employment, that was highly related to taxi ridership.

Origin-destination trip table estimation based on subarea network OD flow and vehicle trajectory data

ABSTRACT Identifying accurate origin-destination (O-D) travel demand is one of the most important and challenging tasks in the transportation planning field. Recently, a wide range of traffic data has been made available. This paper proposes an O-D estimation model using multiple field data. This study takes advantage of emerging technologies – car navigation systems, highway toll collecting systems and link traffic counts – to determine O-D demand. The proposed method is unique since these multiple data are combined to improve the accuracy of O-D estimation for an entire network. We tested our model on a sample network and found great potential for using multiple data as a means of O-D estimation. The errors of a single input data source do not critically affect the model’s overall accuracy, meaning that combining multiple data provides resilience to these errors. It is suggested that the model is a feasible means for more reliable O-D estimation.

Determination of Representative Path Set from Vehicle Trajectory Samples

AbstractThe representative path set between an origin and destination (OD) pair is useful for various transportation areas such as traffic assignment, travel information service, and traffic management. With the development of telecommunication technology, a car navigation system now collects trajectory-type data that can show the routes generated by drivers. Thus, the travel paths of individual travelers are available for the transportation study. Unfortunately, the availability of individual path data cannot guarantee direct application to the transportation model because it does not give us the significant path set between the OD pairs. Every driver in the real world has different starting and finishing points, yet these points are in a same origin and destination zone. In this study, a new methodology to determine the representative path set between an OD pair based on vehicle trajectory data is developed. This new model is tested with simulation data and compares various results according to clusteri...

Analysis of Lane-by-lane Traffic Flow Characteristics in Korea by Using Multilane Freeway Data

PURPOSES: This study analyzed the lane-by-lane traffic flow characteristics in Korea by using real-world data, including congestion levels, for 2-, 3-, and 4-lane freeways. METHODS : On the basis of a literature review, lane flow and speed characteristics were analyzed using flow measurements and speed ratios. In addition, the effect of congestion levels on traffic flow were visualized using rescaled cumulative plots. RESULTS: Driver behavior varied depending on the congestion level. During free-flow conditions, the lane-use ratio of individual lanes varied largely, whereas during congestion, the ratio was nearly the same for all lanes (i.e., equilibrium). During maximum-flow and congestion conditions, the median lane was used more than the shoulder lane, whereas during all other conditions, the shoulder lane had a higher lane-use ratio. In 3- or 4-lane freeways, the lane-use ratio of the median lane always exceeded 1 and was the highest during free-flow conditions. CONCLUSIONS : The results of the present analysis can be used as an index to predict congestion before a lane is overcapacitated. Moreover, the results can be applied in variable lane guidance systems, such as car navigation systems and variable message displays, to control traffic flow.