BACKGROUND

Current transportation safety research includes development of "collision avoidance warning systems" for a number of crash situations, including alerting drivers to a potential collision when they are backing up. Although collisions that result from backing maneuvers are typically not as severe as other types of collisions, their conservatively estimated annual frequency is 500,000--including 50,000 injuries annually and 185 fatalities. These rates make them "a significant safety and cost problem." Teenage drivers and drivers 75 and older have a "disproportionately high" involvement in these collisions.

To be effective, a backup warning system "must capture the attention of an unalerted driver about the presence of an unexpected, unseen, or misperceived object behind the vehicle and result in a timely vehicle-control response." However, certain situations "complicate the implementation of an effective warning system." For example, drivers often intentionally back close to objects; therefore, "because the system does not 'know' whether the driver is aware of the object, there is a high potential for nuisance warnings." In addition, drivers might not see something between them and another visible object. To be effective, however, a warning system must distinguish between the two. The system must also cover different backing situations that occur at a range of speeds.

RECENT STUDY

A recent study defined the recommended characteristics of and developed guidelines for backup warning devices. Neil D. Lerner, Jeffrey L. Harpster, Richard W. Huey, and Geoffrey V. Steinberg reported the study's results in "Driver Backing-Behavior Research: Implications for Backup Warning Devices" (Transportation Research Record 1573). Based on time to collision (TTC), their research showed that a primarily acoustic, two-stage warning system seemed feasible. The researchers also identified topics for further study.

METHODS

This study involved three experiments using licensed drivers equally representative of age groups and gender. The first experiment described normal driving behavior during a variety of common backing scenarios. Participants drove their own vehicles (mostly passenger cars) on a suburban route and were not aware that the study's purpose was to elicit and record backing behaviors. An researcher in the back seat gave motorists directions for driving the route, mixed with instructions to perform backing maneuvers in eight situations. An instrumentation system temporarily installed in each participant's vehicle measured and recorded data.

The second experiment determined "backing driver reaction times and stopping distances in response to a warning signal." Participants drove their own vehicles over a route that included a number of backing scenarios. A temporary, on-board instrumentation system again collected and recorded data. Equipment also included a small button that the experimenter pushed to sound an alarm during backing maneuvers. When drivers heard the alarm (which sounded until the vehicle stopped), they were to stop their cars as quickly and safely as possible.

The third experiment measured "driver perceptions regarding the appropriate presentation times and distances for backing warnings." Such perceptions are significant since "public acceptance will be maximized if the warnings are perceived as valid, timely, and appropriate." To do the experiment, a life-size crash dummy was suspended from the ceiling of an unused level of a public parking garage. One experimenter drove the vehicle, while another sat in the back seat and controlled the data collection equipment. The participant sat in the front passenger seat, looking over his or her left shoulder out the car's rear window. The vehicle backed toward the dummy and hit it; participants were asked to push a button at the moment they thought a warning would be appropriate.

Participants judged appropriate times for both cautionary and danger warnings. The cautionary alarm alerts the driver to a potential collision situation, and the danger alarm means a collision is imminent and the driver must stop immediately. The distinction between the two alarm types was a significant factor. Before the experiment, participants were asked to explain this distinction to confirm they understood the difference between the two types of warnings.

RESULTS

Findings from the first experiment showed that drivers glanced in a number of directions while backing up. Over the right shoulder was the most frequent location; however, older drivers did this less often than younger drivers. Although about 10 percent of drivers looked forward while backing, they did not look at the dashboard. The variety of glance locations and the absence of dashboard glances suggested that any visual warning system should be accompanied by an acoustic warning and that the effectiveness of a dashboard-mounted visual display would be questionable.

Results also showed that, after an initial speed decrease, drivers tended to maintain their speeds during the remainder of the backing maneuver. Consequently, time to collision (TTC--the time it takes a vehicle to hit an object from its current position at its current rate of speed) appears to be a reliable indicator for designing a warning system. Based on TTC, a warning occurring in the 1.5- to 2.0-second range would be reasonable.

Table 1 illustrates the data from the second experiment on brake reaction times. As indicated, 0.54 seconds was the mean brake reaction time for all drivers in all braking maneuvers. Results also indicated "that the reaction time was importantly influenced by where the driver's foot was located at the moment the signal occurred." Because their backing speeds tended to be slower, older drivers stopped in shorter distances than did younger drivers. Overall, the stopping distances observed were "roughly compatible with the TTC data," which "suggests that if TTC is used as a primary basis for triggering a warning, it will be generally consistent with the ability of the driver to stop before reaching the warned-about object." In this experiment, "the reaction times of the drivers appear[ed] consistent with a TTC of approximately 2.0 sec."

The third experiment corroborated the TTC value of 2.0 seconds and showed that "an imminent crash danger signal at a little less than 2.0 sec" seemed reasonable, since "drivers would apparently view this timing as valid and meaningful." The optimal time for a cautionary warning signal was more difficult to pinpoint, since it is more dependent on backing speed. A cautionary warning signal between 3.0 and 5.0 seconds appeared reasonable, but more research in this area would be helpful.

CONCLUSIONS / RECOMMENDATIONS / IMPLICATIONS

Although the data gathered in this study may prove useful for designers of backup warning devices, results "should be treated as preliminary suggestions" and "best estimates" in this phase of ongoing research on backup warning systems. Theoretically, a backup warning device should include both a cautionary and danger warning signal based on TTC, with the former occurring at about 3.5 seconds before impact and the latter at about 1.5 to 2.0 seconds. The warning system should be primarily acoustic and could be supplemented by a visual display located over the driver's right shoulder.

However, because the study was limited primarily to passenger cars, data from other types of vehicles is still needed. In addition, the distinct differences noted in the backing behaviors of older drivers suggested their needs warrant "explicit consideration in any further research." While the study relied on data from a number of drivers in a variety of backing maneuvers believed to be representative of actual drivers in actual driving conditions, "if actual collision events should turn out frequently to involve drivers who are substantially different than the typical drivers studied here, some adjustment to the recommendations might be required."