To evaluate the system, driver experiments were conducted and the RORSIM vehicle dynamics model was programmed to simulate the lane keeping ability of real drivers, and provide estimates of crash potential with and without LDWS on several types of run-off road crash scenarios.

The lane position models captured the meandering and curve-cutting behavior of drivers in three specific situations: Passenger cars on freeways, passenger cars on county roads, and heavy trucks on freeways. Data for the passenger car and heavy truck driver lane keeping models were collected using instrumented vehicles on public roadways.

The simulation study tested the ability of LDWS by sending simulated drivers along randomly generated test paths. At a predetermined point along the path, the steering wheel angle was held in a fixed position to simulate the driver becoming inattentive. If the LDWS warned the driver with enough time remaining to recover control before leaving the roadway, then the LDWS was deemed successful at avoiding a crash during that test run. The performance of the LDWS under the various potential crash situations in the simulations were applied to the crashes recorded in the 1992 General Estimates System (GES) and Fatality Accident Reporting System (FARS) databases.

The study incorporated the following factors in the estimate of the impact of LDWS on Run-off-Road (ROR) crashes:

• Causal factors - The study noted that LDWS can only be expected to be effective on road departure crashes caused by driver inattention or relinquishing of steering control. In addition, the LDWS was assumed to be effective on only 25 percent of intoxicated drivers.

• Available shoulder - Without sufficient room to maneuver on the shoulder, the effectiveness of a LDWS drops substantially. In addition, cars are easier to maneuver than heavy trucks. The study accounted for the reduced performance on narrow shoulders by combining shoulder width distributions documented in the 1992 GES-FARS crash data with LDWS effectiveness estimates at various shoulder widths to estimate the overall percentage of actual road departure crashes that would likely be prevented by a LDWS.

• LDWS availability - Camera-based LDWS may not be designed to function during certain adverse environmental conditions (e.g., nighttime rain with reflections from oncoming lights, snow-covered roadways, sun very low in sky), where lane boundaries are missing, or at low speeds. The study scaled the effectiveness results for LDWS by factors representing the prevalence of these conditions in the GES/FARS single-vehicle road departure crash data.

RESULTS

The results of the study suggested that LDWS have the potential to reduce road departure crashes in passenger vehicles by approximately 10 percent, and reduce road departure crashes in heavy trucks by approximately 30 percent. These reductions would result in approximately 160,000 fewer crashes and 1,500 fewer fatalities in passenger vehicles each year. In heavy trucks, a 30 percent reduction in run-off-road crashes would result in approximately 9,300 fewer crashes and 96 fewer fatalities each year.

The impacts on heavy trucks were relatively higher than those for passenger cars primarily because trucks have a higher frequency of drowsy related crashes and lower frequency of intoxication related crashes compared to passenger vehicles.