Quantitative testing of a frontal collision warning system for transit buses

Abstract

The effectiveness of a frontal collision warning system (FCWS) depends on the ability of its sensors and software to distinguish threatening driving scenarios and to provide the driver with a warning that leads to a safe (collision avoidance) response. A program of testing that focused on determining the capabilities (and limitations) of the FCWS sensors and signal processing is reported, so that the warning system could be designed to make best use of the capabilities and minimise the impacts of the limitations. The tests were conducted carefully under controlled conditions, using target vehicles and obstacles set up to represent typical hazards encountered in urban and suburban transit bus driving. The movements of the subject vehicle (the bus equipped with the FCWS) and the targets were measured independently for comparison with the measurements obtained from the FCWS sensors so that the sources of errors and delays could be determined. The results of the testing indicated that the longitudinal measurements of target locations were generally good, but even relatively small errors in the lateral position measurements were enough to cause problems with false-positive and -negative alerts. The signal-processing delays of approximately 1 s introduced excessive delays in the generation of alerts to the driver. Fusion of the existing sensor data with additional sensors is recommended for future enhancements to overcome these limitations.