Summary

The need for a drivers' model that integrates a wider range of natural driver activities is important to the traffic engineering and human factors communities. Integration of real traffic behaviors into micro-simulations will increase the accuracy and explanatory power of these models. For human factors engineers, improvements to driver modeling efforts will provide a useful framework by which Intelligent Transportation Systems (ITS) can be evaluated for safety and mobility. The proposed extensions include a means to consider normal lane changing maneuver, driver support emergency management and impaired driving. Where warning or taking control from the driver becomes necessary understanding how the cognitive mechanisms operate and when to provide an aid to the driver will be critical to the research effort required.

The basis of a naturalistic driver model, PADRIC, have been established by PATH through two projects (Caltrans Memorandum Of Understanding 369 &ndash Human Driver Models for SmartAHS and Caltrans Task Order 4222 &ndash Human Driver Model Development). These projects led to the definition of the structure of the model and the implementation of the basic vehicle control procedures. In this project, we propose to pursue the development of the model by increasing the scope of simulation capabilities to lane-change maneuvers and emergency or impaired driving. The capability to detect and avoid collisions is integral to safe driving. Determining the structure and pattern of these driver activities under emergency and impaired conditions is central to the extension of the naturalistic driver model.

Methodology

We propose to address the objective of increasing the potential of the driver model by using three principle methods, namely, modeling, experimental, and verification. In the process of modeling, we propose to integrate strategic, tactical and operational information processing levels within one model. The model will be able to simulate many driving maneuvers such as car-following, lane-changing, emergency maneuvers and impaired drivers. To achieve this objective, we will incorporate reactive and anticipative primary operative modes. The "human-like" information processing produced by this architecture will result in commands which affect the control of a vehicle.

To adequately understand and have data to verify the efficacy of extensions of the driver model, experimental investigations of lane-changing, emergency and impaired conditions are essential research activities. The goal of these experimentations is to provide support for the model design and calibration as well as data to validate the model. We plan to conduct complementary (and convergent) naturalistic data collection with the PATH instrumented Taurus for lane changing. A state-of the-art driving simulator will be used to investigate the effects of impaired driving and emergency management. Driving simulation mitigates risks associated with testing drivers in impaired emergency situations where crashes may result. The outcomes of this project will be an improved simulation tool that can recreate many highway-driving situations with human behavior accurately integrated