Summary

This project is aimed at studying a strategy of reducing the frequency of collisions at railroad crossings. It thus aims to improve safety and efficiency of the transportation system in California. The focus of the project is the modification of a commercially available in-pavement warning signal that was evolved from one originally designed to indicate the presence of pedestrians in a pedestrian crosswalk. The question is whether this technology would be capable of supplying an effective visual barrier at a railroad crossing where no gate exists. The first year of the project has been devoted to selecting a crossing for a test installation, studying the visibility of the signals as they are presently supplied, studying means of improving the visibility of these signals, and preparing means of monitoring vehicle behavior during a test. The next year of the project will include installation of a test facility at a crossing in Kern County, testing a vehicle monitoring system at the crossing, and gathering data.

Rail is an increasingly important component of surface transportation in California. A significant impediment to increased use of rail resources is the large number of collisions (over 3000 annually) at grade crossings. Over 350 persons die each year from this cause with many more injured and consequent damage to the efficiency of this transportation modality. Many such collisions occur at crossings with active motorist warnings (e.g. flashing red lights and bells). This raises the question as to whether crossing collisions can be meaningfully prevented. The present project aims to study a novel active warning system that may have the capability of improving the communication to motorists thereby lessening the number of incursions in front of trains.

The subject of the present study, pavement embedded warning lights both in advance of (amber) and at the rail crossing (red) is an example of infrastructure to driver communication that accomplishes both. It is, in its most primitive state, more sophisticated than existing warnings because it is situated where the driver is looking, on the road. But because it is a modality with the potential to effect infrastructure-to-vehicle communications by means of optical communication technology, it is a good example of a deployable and potentially very useful technology even in its nascent state.

The subject matter is a commercially available and MUTCD compliant in-pavement LED warning system that is visually detectable to motorists. The system includes modules (one per lane or a minimum of two on a one-lane installation) each of which contains six separately ignitable LED pairs. We have begun to explore the range of possible space and time signal ignition configurations to achieve optimal signal detectability. For example, one approach is to ignite all LEDs at once. Another is to flash all LEDs with an on-off pattern. We have begun to test more sophisticated possibilities.

Quantification of the detection performance by human observers as measured in the laboratory is the outcome measure used to define the 'optimum'. During the second year, motorist behavior (appraised using a vehicle monitor installation mounted at the crossing), and field tests of visibility in fog (if conditions allow), will be used to assess the value of the crossing warning system.

In the first year, we designed and will have optimized a warning signal system. In the second year we plan to install an optimized warning system at one of the potential test sites listed in the previous section. Driver behavior will be monitored by an in-ground magnetometer-based vehicle sensing installation in which speed of approach and the point at which braking is initiated will be estimated with and without the warning system deployed. Field tests of visibility in fog and of motorist behavior will be used to assess the value of the crossing warning system. As a given crossing may experience an average of fewer than one collision per year, it would not be expected that we would be able to observe improved collision statistics at a single crossing. Accordingly we shall be interested in the more frequently occurring near misses that vehicle monitoring can reveal. Additionally, we will conduct field tests of warning signal detectability in fog if conditions allow.