Using simulation to develop protocols for bicycle crash-avoidance testing.

Abstract

Objective: In the U.S., bicyclist fatalities have risen 47.5% over the last decade. On some of their latest vehicles, automakers have introduced bicycle-detecting automatic emergency braking (AEB) systems that automatically apply the brakes to avoid or mitigate collisions with bicyclists. These systems are not evaluated in the U.S. market, although similar tests are conducted elsewhere. The purpose of this study was to use simulation to understand the AEB system characteristics that might perform well in potential testing protocols.

Methods: Using openPASS, a bicycle and passenger vehicle were simulated traversing through a four-way intersection of two- lane roadways. Both a straight crossing path and a parallel path scenario were simulated with the subject vehicle traveling between 20 and 80 km/h and the bicycle traveling between 5 and 20 km/h. The subject vehicle's sensor field of view (30, 60, 90, 120, 150, 180 degrees) and range (10, 20, 30, 40, 50, 60 m) were varied, and the AEB response was designed to match the braking characteristics observed in pedestrian crash-avoidance testing. In total, 30 hypothetical AEB systems were tested in 20 unique straight crossing path scenarios and 18 hypothetical AEB systems were tested in 24 unique parallel path scenarios.

Results: In the straight crossing path scenario, when evaluating based on avoidance, the simulations where the subject vehicle and bicycle were moving at similar speeds differentiated systems by the sensor field of view. In both straight crossing path and parallel path scenarios, collision avoidance at higher relative speeds was differentiated by the sensor range.

Conclusions: A straight crossing path protocol with the subject vehicle and bicycle moving at similar, low speeds could lead to bicycle-detecting AEB implementations with a wider field of view. The test speed in both scenarios primarily influenced the sensor range. This research provides testing agencies with information about how testing protocol decisions could influence AEB system design. In addition, this study demonstrates the feasibility of using simulation tools to develop relevant crash avoidance testing protocols. Future simulations could predict the performance in real-world bicycle crashes of systems that would also perform well in the potential testing protocols.