Safety

Emergent Collision Avoidance Systems (CAS's) are beginning to assist drivers in performing specific tasks and extending the limits of driver's perception. The systems face significant hurdles in terms of safety and driver acceptance. In fact, CAS's may actually make the driving environment more dangerous. This paper demonstrates that adaptive capability is necessary to mitigate these concerns. Specifically, CAS's must adapt to a driver's style and limitations. Fortunately, a significant portion of this adaptive capability can be realized without additional sensors and with the inclusion of relatively simple hardware. The requirements of components of a CAS are discussed based on advances found in recent relevant literature. An architecture for an adaptive CAS is proposed.

The George Washington University - GW Transportation Research Institute

Presented at the ITS America Annual Conference and Exposition, April 29 –May 2, 2002 Long Beach, California

Traditional ways of directly measuring visual demand while driving (i.e., extracting eye glance data by manual video tape reduction) are tedious and cumbersome. This paper is a summary of the results on visual demand associated with various telematic tasks. The results are based on the data collected from an empirical study. More specifically, traditional eye glance measures such as Mean Single Glance Time, Number of Glances, Total Glance Time, and Longest Single Eye Glance are compared to other dependent measures such as Static/Dynamic Task Completion Times and the Number of Excursions committed while completing the task. Previous findings indicated that static task completion time is a viable surrogate of dynamic task completion times. Results presented herein provide supportive evidence of previous findings, in addition to the Total Glance Time being a good predictor of other surrogate measures such as lanekeeping performance and task completion times.

Ford Research Laboratory

Virginia Tech Transportation Institute

Presented at the ITS America Annual Conference and Exposition, April 29 –May 2, 2002 Long Beach, California

Emergency Evacuation is one most protective measure and viable alternative during regional emergencies in response to both natural and man-made disasters. Several evacuation traffic models have been developed and currently available to support the planning and analysis of emergency evacuation. However, to be effective, the decision-makers must understand how these models can be used to facilitate the planning, analysis, and deployment of emergency evacuation for populations at risk. This paper reviewed and analyzed various traffic models, suggested how to improve the operational planning of emergency evacuation, and recommended the necessary technological enhancements for evacuation traffic models.

RSPA/US.DOT

Oak Ridge National Laboratory

Presented at the ITS America Annual Conference and Exposition, April 29 –May 2, 2002 Long Beach, California

As an example of the newly established ITS Archive Data User Service (ADUS), this study utilizes information technologies to archive incident data collected by NAVIGATOR, Georgia’s ITS system, which makes it feasible to incorporate sophisticated speed contour and incident mappings to unveil the possible causal factors of secondary accidents. This study brought out a new definition of secondary accidents: an accident is a secondary accident if it falls in the congested contour area caused by the initial incident. Preliminary findings in this study demonstrated the suitability and usability of such a definition. State Department of Transportation and local transportation agencies could exploit the possible benefit of eliminating all the secondary accidents to enhance the support for the deployment of Intelligent Transportation System.

Georgia Institute of Technology

Presented at the ITS America Annual Conference and Exposition, April 29 –May 2, 2002 Long Beach, California

Presentation

• Need

– For forward collision warning applications, need accurate estimation of host vehicle forward path and state to determine in-path vs. out-of-path vehicles

– Single sensor systems (gyro, vision, GPS, radar) have high error and drop-out rates due to sensor limitations and degradations (weather, visibility, traffic density, road state, buildings, etc.)

– Multi-sensor systems could potentially increase accuracy, but need smart fusion methods to handle individual sensor drop-outs and incomplete/redundant/contradictory outputs

• Objectives

– Develop algorithms to fuse multiple sensors to improve accuracy and reliability of host vehicle forward path and state (as part of ACAS-FOT program)

• Status

– Multi-sensor fusion module completed and integrated into ACAS-FOT prototype vehicle (Phase 1 testing complete)

HRL Laboratories, LLC

Presented at the ITS America Annual Conference and Exposition, April 29 –May 2, 2002 Long Beach, California