Agent-based modeling and simulation in the dilemma zone / Sojung Kim [and four others].

Format:

Book

Government document

Author/Creator:

Kim, Sojung, author.

Contributor:

United States. Federal Highway Administration. Office of Research, Development, and Technology, issuing body.

United States. Federal Highway Administration. Office of Safety R&D, sponsoring body.

Language:

English

Subjects (All):

Exploratory Advanced Research Program (U.S.).

Roads--Interchanges and intersections.

Roads.

Motor vehicle driving--Decision making.

Motor vehicle driving.

Physical Description:

1 online resource ([335] pages) : color illustrations

Other Title:

At head of title: VASTO-Evolutionary Agent System for Transportation Outlook

Place of Publication:

McLean, VA : U.S. Department of Transportation, Federal Highway Administration, Research, Development and Technology, Turner-Fairbank Highway Research Center, December 2015.

Summary:

The goal of this study is to develop a realistic dilemma zone (DZ) model that considers the effects of factors surrounding vehicles at an intersection, particularly focusing on driver decisionmaking behavior, such as the presence of a pedestrian countdown signal, a red-light photo enforcement camera, and the actions of an adjacent vehicle. The Federal Highway Administration's Highway Driving Simulator (HDS) was used to collect drivers' responses as a function of facility speed limit and whether the driver was in a hurry. From these responses, a DZ model was developed via agent-based modeling and simulation (ABMS) under the extended belief-desire-intention (E-BDI) framework. This framework represents uncertain perception and decision behaviors of humans in a probabilistic manner. To demonstrate and validate the proposed approach, the drivers' actions under two circumstances were compared: (1) drivers only knew the approach speed and distance to the stop (i.e., internal information); and (2) drivers knew the internal information as well as external information, such as presence of a pedestrian countdown signal or presence of a redlight photo enforcement camera, and the behavior of an adjacent vehicle. The experiments indicate that sufficient external information could accurately predict the decisions of drivers and thus reduce the number of red-light violations that could potentially cause accidents at an intersection. In addition, the proposed E-BDI-based DZ model allowed us to consider the physical interactions between vehicles such that it provides realistic prediction results of drivers under various traffic conditions.

Notes:

Title from title screen (viewed Feb. 5, 2016).

"December 2015."

"FHWA-HRT-15-082."

Includes bibliographical references.

OCLC:

937708567