After a decade of reductions in passenger fatalities by improving vehicle crash safety, pedestrians now account for the majority of traffic accident fatalities in Japan. Collision Avoidance and Mitigation Systems (CAMS) are intended to monitor objects ahead including pedestrians, issue a warning to the driver upon detecting an object, and activate automatic brakes. CAMS are promising technologies for reducing pedestrian and motor vehicle accidents. However, there are currently no standardized test methods for evaluating their safety performance and they have been slow to spread in the market. This study proposes a protocol for evaluating the performance of CAMS, and estimates their effect on reducing pedestrian fatalities and injuries.
We used two test vehicles with CAMS having different sensing systems. To investigate the collision avoidance performance of CAMS, a test vehicle was driven toward a pedestrian dummy which was set up on a test course, and the collision avoidance situations were recorded. Among various test conditions, daytime, dry road surface, side-facing pedestrian, black clothing (pedestrian), and center position (of the vehicle) were selected as standard test conditions.
In evaluating the performance of CAMS, we used the criterion of whether or not a collision with the pedestrian dummy was avoided without any operation by the driver. The results showed substantial variability in collision and avoidance even under the same standard conditions. In order to include the uncertainty of the collision avoidance results, we assumed collisions to be probabilistic events. By applying a logistic regression model with “p” as the probability of pedestrian dummy collision and vehicle speed “x” as an explanatory variable when using CAMS under the standard conditions, we defined collision probability “p(x)” as the performance of CAMS. p(x) clearly shows the differences in performance between two vehicles tested.
We analyzed factors contributing to the differences in performance. As the two main functions of CAMS are to detect pedestrians and to apply the automatic brakes, we used the warning timing as a measurement of the detection function, and the braking timing as a measurement of the automatic brake function. An analysis of the difference in collision avoidance performance between the two vehicle models showed that the timing of automatic brake activation is the cause of the difference. It was also found that in order to increase the collision avoidance probability, it is more effective to activate the automatic brake based on CAMS’ judgment, rather than to wait for the driver to respond to a warning.
In the traffic fatality and injury data, we estimated the fatality reduction effect of CAMS by applying the defined accident avoidance probability of CAMS. Due to the performance of CAMS, the effect on reducing pedestrian fatalities is larger at low and medium speeds. CAMS also have a more significant effect on reducing severe injuries because the rate of severe injuries is higher at low and medium speeds where the CAMS collision avoidance probability is higher.