Contact Us       Opportunities       Hawkmail

How do I get to NADS?

Publications »
Automatic Learning by Autonomous Driver Agents as Applied to Performing Realistic Lane Change Maneuvers

Document Number:N2001-012
Document Type:Conference paper
Author(s):O. Ahmad
Y. Papelis
S. Bulusu
V. Gade
Publication / Venue Name:ISHF2001
Publication Date:2001-09-21
Abstract:High-fidelity driving simulators immerse a driver in a highly realistic virtual environment for the purpose of studying human driving behavior in a realistic yet safe setting. Many factors contribute to the immersive experience, but when the subject under study requires interaction with other vehicles, it is important that the virtual environment includes a microscopic traffic simulation model. Such a model often consists of populating the road network with one or more autonomous driver models. The more individual types of maneuvers an autonomous driver model exhibits, the more realistic it appears to the simulator driver. This paper focuses on the design of a specific behavior that is responsible for implementing lane change maneuvers. This behavior, which is part of a larger autonomous driver model, is important because it is necessary for building more complicated behaviors such as yielding, route planning, and merging. This behavior is unique in that it accommodates several conditions simultaneously and is designed to work with an arbitrary vehicle dynamic model. To address the variability in the physical response of different vehicle models, it uses an off-line learning technique that exercises the dynamic model and builds lookup tables that encompass the experience of the driver model.
Keywords:None listed

Bookmark and Share
Like us on Facebook!
Coypright © 2014 - The National Advanced Driving Simulator