Distribution of renewal cycle under varying conditions

Table 4-3 shows the distribution of the common renewal cycles under various attributes. There were ten types of renewal cycles with more than a 1% frequency share, whereas eight types of 2-glance renewal cycles accounted for 90.74% of the frequency. Among the 2-glance renewal cycles, those involving in-vehicle distractions and rear-view mirror glances accounted for almost half of the generated renewal cycles.

To analyze the characteristic differences among various attributes, the distribution of these common renewal cycles under different conditions was examined.

The recorded maneuvers and their relation to a junction were referred to the final pronounced action and associated location before a precipitating event. Such attributes did not necessarily exist throughout the entire 30-s data period. Certainly, the generated renewal cycles might occur before or during the maneuvering. Thus, it seems that a mismatch of time exists between eye-glance data and certain driving circumstances. Nevertheless, before implementing maneuvering intentions, drivers tend to look in the directions of future vehicle trajectories. That is, the entire maneuver includes searching for information, decision-making, and the final action.

Analyzing only the exact period of the maneuver does not represent the entire attention allocation process. Thus, from this view point, the mismatch problem is ignored.

The attributes of a relation to a junction and maneuver were important for determining a driver’s expectations of potential threats. In these cases, of a relation to a junction, road segments and intersections were the two main elements in the driving environment. When the drivers encountered intersections within 30 s, more renewal cycles of RF and RW would occur, probably because of the associated possibility of increased conflicts from the intersected roadway.

Table 4-3 Distribution of renewal cycles by attributes

Attributes

Sample size

Distribution of renewal cycles (%)

F-InvD F-ReM F-LW F-LM F-RF F-RW F-LF F-RM F-LM-LW F-RF-RW Others

Pre-event maneuver

* The renewal cycles related to directions of turning and changing lane were only recorded while the turning signal was turned on.

Lane changing and turning were the two primary maneuvers that naturally directed the drivers’ attention to directions critical for preventing conflicts. Meanwhile, drivers decreased the attention invested in non-safety related areas, such as InvD. The percentage of the renewal cycles in which the drivers transited attention to InvD decreased from 26.4% when driving straight to 12.4% when changing lanes, and to 17.4% when turning left or right. While changing lanes, the sample drivers increased their attention to the ReM and LM to observe the traffic conditions behind them.

In particular, the drivers transited their vision more frequently to the left side (LW and LM) when changing to the left lane, and to the right side (RF and RW) when changing to the right lane. The main difference between changing to left or right lanes was the use of the side mirrors. The RM was seldom used when changing to the right lane. One reason might be the faster driving speed in the inner (left) lane. Vehicles located in the right rear area were usually traveling relatively slowly. Once the drivers had successfully passed those vehicles, they had good information for where the

vehicle was and could easily begin changing lanes to the right side without glancing at the RM. By contrast, vehicles in the left lane usually traveled more rapidly and required drivers’ close attention to ensure a safe margin for changing lanes to the left.

Turning at intersections was indicated to have an increased risk of crashing with traffic from an intersecting roadway in front of the subject vehicles. An increased number of renewal cycles involving LF, RF, and LW glances were found. For maneuvering a left turn, the LW, LF, and RF were the most common focal points for checking the potential threats coming from opposite traffic. Glancing at those focal points implied that threats were expected from the traffic passing through intersections. Another unique characteristic of turning left was the high percentage of the path from Forward to InvD (F-InvD). Turning was usually associated with complex tasks and few chances of shifting one’s attention to InvD. However, the drivers were more likely to stop and wait at the intersection when turning left than when turning right and changing lanes. In the absence of immediate crash risks, drivers may be inclined to use in-vehicle devices or interact with passengers while waiting. For a right turn, more potential conflicts were related to the traffic from the intersected roadway, pedestrians on the crosswalk, and cars following behind. Thus, the sample drivers paid greater attention to monitoring the ReM, LW, and RW.

Traffic density determined interactions with other vehicles. When traffic density increases from Level of Service (LOS) A to D, the sample drivers allocated more attention to the ReM and LM, probably checking traffic from behind or for lane changing. Moreover, the necessity for frequent speed adjustments and the shorter available reaction time associated with heavy traffic discourages drivers from engaging in non-driving-related tasks, such as transiting their vision from the forward areas to the roadside areas (LW, RW, and RF) or attending to InvD. When traffic density increased to LOS E, the sample drivers were unable to operate their vehicles freely but were forced to remain in the traffic stream. Under such conditions, drivers had ample opportunities to use in-vehicle devices because of the slow traveling speed and limited gaps available to merge with other vehicles. Thus, the percentage of InvD climbed sharply from 19.0% under LOS D to 29.9% under LOS E.

Among the common cycles, InvD were the main focal points on which the drivers spent a large portion of their non-forward attention time. As shown in Table 3, when distractions were present, F-InvD contributed 45.3% of the extracted renewal cycles. However, in the absence of distracting activities, 16.8% of the renewal cycles were still related to F-InvD. These findings suggest that engaging in distracting activity was not the only reason that the drivers transitioned their vision to in-vehicle focal points. At times, drivers transited vision inside their vehicles, despite doing

nothing with in-vehicle devices. Because the sample drivers represented by this data set eventually experienced crashes or near crashes, it is reasonable to presume that defective behavior might have occurred in their daily driving operations.