We at the Institute for Traffic Accident Research and Data Analysis (ITARDA) have studied the particulars of about 3,000 traffic accident cases over the past ten years. From these cases, we selected 152 cases of rear-end collisions between motor vehicles and analyzed the errors committed by the drivers of the rear-ending vehicles in those accidents. (1)Driving is a continuous process of perception, decision-making/prediction, and operation/action.   The ability to navigate safely through the flow of traffic requires a continuous process on the part of the driver from perception to decision-making/prediction to operation/action, as charted in Figure 6. Perception involves not only seeing but perceiving objects that must be grasped in order to drive safely. Decision-making/prediction involves such processes as determining whether the perceived object is stationary or moving, or predicting another object behind the perceived object, and then figuring out what operation/action should be taken. Lastly, the process is completed by operation/action. In other words, driver errors may occur in three stages.   Also, once the driver has made an error in perceiving an object (failure to see an object, etc.), it precludes the next part of the process (decision-making/prediction) from occurring.   Quite simply, the work of perception can be called the portal to safe driving.

(2)A casual attitude toward decision-making/prediction and operation/action during normal driving promotes accidents.   The analysis results indicate that in many accidents a casual attitude toward decision-making/prediction and operation/action during normal driving promotes errors immediately before an accident, that is, "direct errors." "Background errors" is the name we gave to the errors that foster such direct errors. In other words, errors can be divided into two types according to time sequence.   This analysis used error data only up to the point where accidents were considered avoidable if the proper perception, decision-making/prediction, and operation/action had been taken at the time; cases of panic reactions beyond that point or unintentional actions were omitted from the analysis. (3)About 2.6 errors are committed per person in one accident.   Figure 7 indicates the number of errors by type of error, divided into "direct errors," which occur just prior to the rear-end collision, and "background errors," the preceding situations that induce drivers to be inattentive. The figure shows that a total of 269 errors were committed just before the accidents, and about 80% of these were perception errors. On the other hand, there were 122 background errors in total and about 80% of these were decision-making/prediction errors. If we total the "background errors" and "direct errors," we get 391 error cases, which means that the drivers of the rear-ending vehicles committed about 2.6 errors per person (391 cases/152 drivers). In other words, the drivers had two to three chances of avoiding an accident.

(4)Driver not looking ahead just before accident occurs; Analysis of direct errors   We analyzed perception errors, which constitute the bulk of "direct errors." In Figure 8, the vertical axis represents objects of perception error, and the numbers of perception errors are shown in different colors according to their factors.

1)Objects of perception error   In about 90% of all cases, the object of perception error is a preceding vehicle or a parked, stopped, or disabled vehicle, which would eventually be rear-ended.   Other objects of perception error include "Other vehicle's signal" (in this case the driver of the rear-ended vehicle is at fault), which causes a misunderstanding on the part of the rear-ending vehicle's driver. Such errors, which account for about 6% of all cases, typically occur when the driver of the rear-ended vehicle is slow to signal a turn or apply the brakes, causing the driver of the rear-ending vehicle to commit a perception error. The "Other" category includes cases in which a driver commits an error in perceiving the traffic signal and rear-ends another vehicle. 2)Factors in perception error   From Figure 8, it is evident that the cause of perception errors can be summed up by "I did not see the object, although I could have" (represented by "Did not see ｡ﾄ" in Figure 8). Another cause that is heard occasionally is, "I could not see the object even if I had tried" (represented by "Could not see ｡ﾄ" in Figure 8). For example, a driver fails to see a parked, stopped, or disabled vehicle on the road shoulder or roadway "because it is too dark" or "because the car has no lights on." 3)Breakdown of "Did not see ｡ﾄ"   As is apparent from Figure 9, in the majority of cases the cause of "Did not see ｡ﾄ" is "I was focused on something else (inside or outside the vehicle) and was looking aside." About one-fourth of all cases are due to "drowsiness, drinking, sudden illness." Although there is no justification for drinking, drowsiness and illness are something that can be avoided by taking care of one's health.

(5)Factors that induce drivers to be inattentive   Having explained the causes of errors that occur just prior to accidents, we are left wondering what could so simply divert drivers' eyes (attention) from looking ahead.   Accordingly, we will next consider the factors that induce drivers to be inattentive. 1)Factors that induce drivers to be inattentive (1): "Background Errors"   During normal driving, decision-making/prediction can easily cause seeing what one wants to see (wish).   We know that 65 of the 152 drivers in total committed errors during normal driving when they were not predicting the danger of an accident. And about 80% of those errors were decision-making/prediction errors.   Almost 80% of all errors involved inappropriate decision-making/ prediction regarding objects that had been correctly perceived.   The objects for which decision-making/prediction errors occurred despite correct perception were: the preceding vehicle (57%), choice of the speed of one's own vehicle (23%), and traffic signals (10%).   As was previously explained, the preceding vehicle is most frequently perceived in error due to lack of attention just before a rear-end collision. The fact that drivers make perception errors for objects they have firmly perceived just a while ago (during normal driving) has a significant meaning.   What were the drivers who perceived these objects thinking or what were they predicting? Using the preceding vehicle, the most frequent object of perception error, as an example, we arranged the results of our findings in Figure 10.

The vertical axis shows the contents of the decision-making/prediction errors, and the factors are color-coded and expressed in specific terms in the legend. The contents of the decision-making/prediction errors include: "Preceding vehicle will not decelerate/turn left or right" and "Distance between vehicles is appropriate." In cases where the preceding vehicle is stopped at a traffic signal or in heavy traffic, the contents include: "Preceding vehicle is farther away" and "Preceding vehicle will start to go forward immediately." Of course, it might be more appropriate to call these contents expectation/wish, rather than decision-making/prediction.

2)Factors that induce drivers to be inattentive (2): Road and Traffic Conditions   In 1-(4), we explained that most rear-end collisions occur at non-intersection locations, where there are no traffic signals and the flow of traffic should be relatively smooth. Of course, the primary reason for this is that most roads are not intersections and have no traffic signals, but other reasons can also be considered. That is, in the vicinity of intersections where traffic is complicated, drivers naturally predict that the preceding vehicle will slow down or stop. On the other hand, when driving at monotonous (or seemingly monotonous) non-intersection locations, drivers will think that there is no reason for the preceding vehicle to decelerate or stop; so many trailing vehicle drivers probably give themselves "some degree of latitude or inattention" in the way they drive. In reality, however, there are openings in center strips (to allow vehicles to turn right or make a U-turn), on and off ramps for roadside services, and other "junction points that are hard to see." In these areas, there are many vehicles turning left or right or decelerating or stopping. What is more, the flow of traffic is often disrupted by parked vehicles, construction, accidents, traffic jams, and so on.   We therefore classified the locations (conditions) that gave rise to rear-end collisions according to whether the preceding vehicle's decelerating or stopping actions were "easy to predict" or "difficult to predict," and then compared the frequency of accidents. The results of this comparison are shown in Figure 11.

A summary of these results indicates that the rear-end collisions in which the actions of the preceding vehicle are difficult to predict - namely, "Hard-to-see junction point," "Tail end of traffic jam," "Construction/accident, and "Parked or stopped vehicle" - account for 76 of the 152 cases in total. In other words, we were able to confirm that even at non-intersection locations the flow of traffic can be disrupted, producing a large number of rear-end collisions.   Although there were only five rear-end collisions that occurred "In heavy traffic," we expect such accidents to become more frequent if we conduct a national survey, judging from the fact that traffic jams occur with a bias toward time and place.   There were 21 cases of a vehicle rear-ending a preceding vehicle in motion, but 16 of them occurred on highways, probably because the monotony of highway driving causes drivers to drop their guard. Typical rear-end collision patterns are diagramed in the following. The background errors are shown in [___](Figure 12).

UP

Institute for Traffic Accident Research and Data Analysis (ITARDA)