U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

 
REPORT
This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-17-024    Date:  June 2017
Publication Number: FHWA-HRT-17-024
Date: June 2017

 

Cooperative Adaptive Cruise Control Human Factors Study: Experiment 4—Preferred Following Distance and Performance in An Emergency Event

CHAPTER 3. PART 2

EQUIPMENT

The Driving Simulator

The same driving miniSim™ driving simulator setup as in part 1 was used.

Multifunction Display

A 17.78-cm (7-inch) (11.43- by 19.05-cm (4.5- by 7.5-inch)) liquid crystal display (LCD) touch-screen display was mounted just to the right of the steering wheel in the driving simulator, which was similar to where the center console would be in a full cab vehicle. The display was used to turn on the CACC system at the beginning of the drive and if the driver used the brake. For all conditions, the screen displayed the vehicle’s set speed (always set to 112.7 km/h (70 mi/h)), the following distance (always set to near), and the status of the CACC system (engaged or not engaged). The engage button on the right side of the display could be used by the participants to engage CACC. When the system was engaged, the text and icons appeared green; when the system was not engaged, the text and icons appeared red.

SIMULATION SCENARIOS

The same simulated eight-lane interstate highway was used in both parts 1 and 2 of the study. Participants in part 2 completed the same drives as the participants in part 1. In addition to these two drives, participants completed a third drive. The third scenario used the same roadway as the first two drives.

The participants began stopped on an on-ramp in the third position of a four-car platoon. As the scenario began and the ramp meter turned green, the platoon proceeded down the ramp and accelerated to 112.65 km/h (70 mi/h) while maintaining the appropriate gap (0.6 or 1.1 s (described in more detail later)). Approximately 5 min into the drive, another CACC vehicle merged into the platoon directly in front of the participant and halfway (9.33 or 17.07 m (30.6 or 56 ft)) between the participant’s vehicle and the vehicle the participant had been following. The CACC system adjusted the gaps of the affected vehicles back to the assigned gap. If the participant braked in this situation, then the CACC system disengaged and needed to be reengaged.

Approximately 20 min into the drive, a vehicle sped down an on-ramp, merged in front of the platoon, and crashed. (The crash was animated but was not in the participant’s line of sight if driving centered in the travel lane.) The crash avoidance event began when the lead vehicle in the platoon decelerated at 9.75 m/s2 (32 ft/s2). One tenth of a second after the lead vehicle began braking, all of the CACC vehicles behind it simultaneously began to decelerate at 3.90 m/s2 (12.8 ft/s2) and illuminated their brake lights. The engine noise was set up to exaggerate the engine revolutions per minute to more adequately cue 3.90 m/s2 (12.8 ft/s2) deceleration.

GAP ASSIGNMENT

Participant assignment to gap groups was based on mean gap maintained during the second drive. Based on the median following distance of part 1 (0.91 s), participants’ preferred following distance was determined. Participants were assigned as near or far preference group based on whether their preferred gap fell above or below the median. Next, participants were assigned to either a congruent or incongruent following preference group (see table 3). Those participants who completed the second drive with a mean following gap of less than 0.91 s were determined to have a preferred near following distance. Of the 59 drivers with a preferred near following distance, 29 were assigned to the congruent (0.6 s near) gap in drive 3, and 30 were assigned to the incongruent (1.1 s far) gap in drive 3. Those participants who completed the second drive with a mean following gap greater than 0.91 s were determined to have a preferred far following distance. Of the 39 drivers with a preferred far following distance, 20 were assigned to the congruent (1.1 s far) gap in drive 3, and 19 were assigned to the incongruent (0.6 s near) gap in drive 3.

Table 3. Participant following distance group assignments.

Group
Assignment
Gap Following
Distance (Drive 2) (s)
Assigned Following
Distance (Drive 3) (s)
Congruent < 0.91 (Near) 0.6 (Near)
Incongruent < 0.91 (Near) 1.1 (Far)
Incongruent > 0.91 (Far) 0.6 (Near)
Congruent > 0.91 (Far) 1.1 (Far)

 

Gap following distances were selected based on quartiles 1 and 3 following distances during the close drive. A 0.6-s gap (quartile 1) was used as the near following distance, and a 1.1-s gap (quartile 3) was used as the far following distance.

WORKLOAD ASSESSMENT

Driver workload was assessed by administration of the National Aeronautics and Space Administration Task Load Index (NASA-TLX) and was measured three times.(9) The first assessment was approximately 5 min into the drive. It was administered after the vehicle merged in front of the participant at about 30 s after the platoon had stabilized. The second assessment was administered approximately 10 min into the drive and was intended to assess the workload associated with driving in a stable, unchanging state (i.e., a baseline index). At this point, participants were between merging events and were likely to feel comfortable with the driving task in general. The third and final NASA-TLX was administered immediately after the final collision avoidance event.

PARTICIPANTS

Participants included 98 licensed drivers recruited from the Washington, DC, metropolitan area. Participants were required to be at least 18 years of age and were screened for susceptibility to motion and simulator sickness. Out of the 98 participants, 49 were male and 49 were female. The participants’ ages ranged from 21 to 73 years with a mean age of 43.3 years (median 43.5 years). Roughly equal numbers of participants under and over the age of 45 were recruited (see table 4). Data from five participants were excluded from analyses because of corrupt data files.

Table 4. Total number of participants included in each condition.

Preferred
Gap
Assigned Gap Total
Near Far
Near 27 (14 males) 30 (15 males) 57 (29 males)
Far 17 (9 males) 19 (9 males) 36 (18 males)
Total 44 (23 males) 49 (24 males) 93 (47 males)

 

PROCEDURE

Participants experienced the same welcome and screening and were provided with the first two drive procedures as those participants in part 1. After the completion of the second drive and SSQ, a slideshow presentation was shown to all participants. The presentation provided an overview of the experimental instructions and familiarized participants with the NASA-TLX questions. The participants assigned to the CACC conditions were also shown videos that explained the CACC concept.

After being provided with a time to answer questions and clarify the use and functionality of the CACC system, participants were escorted back to the driving simulator. The use of the multifunction LCD display was explained and demonstrated as necessary. Next, participants drove the third scenario while using the CACC system. If at any time the participants used the brake or otherwise disengaged the CACC, then they were reminded to use the multifunction display to reengage the system.

After the completion of the third drive, participants completed a second SSQ. They were given time to ask any questions about the study, debriefed, thanked, and paid for their time. In total, participation lasted 60–90 min.

RESULTS

Drives 1 and 2

Participants in part 2 completed the same two drives as those participants in part 1. The data from the first comfortable following task were not used to determine following distance preference; those data are presented here for descriptive purposes. To provide participants with sufficient time to adjust following gap for each speed change (88.5, 104.6, 112.7, and 88.5 km/h (55, 65, 70, and 55 mi/h)), the first 30 s of vehicle following at each speed were excluded from analysis. Table 5 presents drivers’ following time gap distributions by speed averaged across participants during the comfortable following distance drive.

Table 5. Participant following time gaps (s) by speed during comfortable following drive in part 2.

Speed
(km/h)
Minimum
(s)
Quartile 1
(s)
Quartile 2
(s)
Quartile 3
(s)
Maximum
(s)
Mean
(s)
88.5 0.69 1.63 2.17 2.80 6.66 2.33
104.6 0.62 1.54 2.15 3.09 21.90 2.82
112.7 0.41 1.57 2.21 3.08 21.04 2.94
88.5 0.34 1.51 1.83 2.57 7.44 2.18
1 km/h = 0.62 mi/h

Data from the second drive were used to classify drivers as having a near or far following distance preference. Once again, in order to provide participants with sufficient time to adjust following gap for each speed change (88.5, 104.6, 112.7, and 88.5 km/hr; 55, 65, 70, 55 mi/hr), the first 30 s of vehicle following at each speed were excluded from analysis. Table 6 presents drivers’ following time gap distributions by speed averaged across all participants during the comfortable following distance drive.

Table 6. Participant following time gaps (s) by speed during close following drive in part 2.

Speed
(km/h)
Minimum
(s)
Quartile 1
(s)
Quartile 2
(s)
Quartile 3
(s)
Maximum
(s)
Mean
(s)
88.5 0.30 0.64 0.79 1.12 3.96 0.99
104.6 0.27 0.59 0.80 1.18 5.25 1.06
112.7 0.28 0.58 0.82 1.20 6.93 1.08
88.5 0.35 0.61 0.90 1.27 14.61 1.29
Mean 0.30 0.60 0.83 1.19 7.69 1.10
1 km/h = 0.62 mi/h

 

The median following distance for the participants in part 2 dropped 0.08 s from 0.91 to 0.83 s. However, the overall distribution of scores remained relatively consistent.

Workload

The NASA-TLX was administered verbally at three points during the third drive: shortly after the first merge, during a cruise period, and after the final crash event. The effects of CACC following distance on workload were tested using generalized estimating equations (normal response distribution and identity link function) with NASA-TLX as a repeated measure and experimental treatment conditions (preferred gap and assigned following gap) as the between-group factors of interest (see figure 1).

This figure is a bar graph displaying National Aeronautics and Space Administration Task Load Index (NASA-TLX) values by preferred time gap, assigned time gap, and assessment location. The x-axis shows preferred gap, and the y-axis shows NASA-TLX and ranges from 0 to 100. The values are grouped by assessment location as follows: just after the merge event, during the cruise period, and after the crash event at the end of the drive. Within each group, bars are shown based on both assigned and preferred following gaps. Mean NASA-TLX values after the merge event are as follows: assigned near gap and preferred near gap is 14.58, assigned near gap and preferred far gap is 27.31, assigned far gap and preferred near gap is 9.54, and assigned far gap and preferred far gap is 16.25. Mean NASA-TLX values during the cruise period are as follows: assigned near gap and preferred near gap is 10.69, assigned near gap and preferred far gap is 19.49, assigned far gap and preferred near gap is 10.07, and assigned far gap and preferred far gap is 14.05. Mean NASA-TLX values after the crash event are as follows: assigned near gap and preferred near gap is 53.70, assigned near gap and preferred far gap is 46.78, assigned far gap and preferred near gap is 58.23, and assigned far gap and preferred far gap is 64.56.

Figure 1. Graph. Mean NASA-TLX scores by preferred time gap, assigned time gap, and assessment location.

 

As expected, the location of the NASA-TLX assessment significantly affected perceived workload ( χ2(2) = 129.81, p < 0.001). The mean NASA-TLX score after the first merge (M = 15.76) was not significantly different than during the cruise period (M = 13.11). However, mean workload was significantly greater after the final crash event (M = 55.44) than at the two previous times.

Merging Vehicle Response

To assess trust and comfort in the CACC system, participants’ responses to the merging event was evaluated. After approximately 5 min of driving, a vehicle traveled down an on-ramp and merged directly in front of the participant vehicle. The CACC system was programmed in such a way that it was not necessary to interfere with or override the vehicle speed during the merge. Nonetheless, three older participants (two males and one female) pressed the brake pedal. Table 7 shows participants’ preferred and assigned following distances along with the time until the brake pedal was depressed during the merging event. Time is relative to the merging vehicle entering half way into the travel lane. As such, the participant with the negative value anticipated the vehicle merge and braked in advance. Because only three participants pressed the brake pedal during this merge event, braking was not further analyzed.

Table 7. The time from merging vehicle entrance into lane until brake press.

Preferred
Following
Distance
Assigned
Following
Distance
Time from Merging Vehicle
Entering Travel Lane to Driver
Depressing the Brake Pedal(s)
Far Near 0.01
Far Near −2.00
Far Far 1.20

 

Despite few participants using the brake pedal, the speed and short distance between the participant’s vehicle and the merging vehicle may have made the participants uncomfortable. As a result, foot position immediately prior to and during the vehicle merge was explored.

Foot pedal video data were coded beginning at 30 s immediately preceding the merge event. Each participant’s foot position was noted. Foot movement to the brake pedal into a hovering position in anticipation of the merging vehicle was coded. Those participants following at the near following distance were significantly more likely to hover over the brake pedal than those at the longer distance ( χ2(1) = 5.27, p = 0.022). This is not surprising given the short distance between the participant’s vehicle and the merging vehicle, which was likely to generate some mild discomfort in following distance. Participants appeared to be readying themselves to apply the brake if necessary.

No difference in foot hovering over the brake was found based on preferred following distance ( χ2(1) = 0.29, p > 0.05). Further, no significant interaction between preferred following distance and assigned following distance was found ( χ2(1) = 0.00, p > 0.05). In other words, those participants who preferred to drive at a shorter following distance were just as likely to hover their foot over the brake pedal as those who preferred a longer following distance.

Crash Event Reaction

Given that it is possible that participants may have anticipated a collision event or some other nonrecurring traffic event, participant foot hovering prior to the final crash event was explored. Only three participants hovered their foot over the brake prior to the crash. Lateral position within the lane could also have the potential to influence participant reaction time. However, in this study, lateral position within the lane was not found to significantly affect reaction time (p > .05). Taken together, it is supposed that participants did not anticipate a crash or other abnormal driving event.

All but three of the participants depressed the brake pedal during the final event. Next, participant reaction time for those who used the brake pedal was explored. Participant reaction time was calculated as the time between when the principal other vehicle entered the line of traffic and when the participant first depressed the brake pedal (see figure 2). Participants who drove at the close distance depressed the brake pedal (M = 2.77 s) significantly faster than the participants who drove at the far distance (M = 4.88 s) ( χ2(1) = 14.34, p < 0.001). No difference in reaction time based on preferred following distance was found ( χ2(1) = 0.00, p > 0.05). Similarly, no interaction between preferred and assigned following distance was found ( χ2 (1) = 1.08, p > 0.05).

This figure is a bar graph displaying the mean time from the principal other vehicle that caused the crash event after entering the traffic flow until the participant depressed the brake pedal. The x-axis shows preferred gap, and the y-axis shows time from event to brake onset and ranges from 0 to 6 s in increments of 2 s. The values are grouped by both assigned and preferred following distances (near and far). The mean time values are as follows: assigned near gap and preferred near gap is 2.98, assigned near gap and preferred far gap is 2.23, assigned far gap and preferred near gap is 4.39, and assigned far gap and preferred far gap is 5.35.

Figure 2. Graph. Time (s) from principal other vehicle entering the traffic flow to
participant brake pedal onset based on preferred and assigned time gaps.

 

Another manner in which brake pedal response was examined was through maximum pedal depression. The time from initial brake pedal press until the maximum pedal depression was explored. Preferred following distance approached significance in an interesting way ( χ2(1) = 3.54, p = 0.059). Those participants who preferred to follow at a far distance reached maximum pedal depression (M = 1.28 s) faster than those who preferred to follow at a near following distance (M = 2.05 s). While the difference was not significant, the trend shows that those participants who were more comfortable following at a greater distance may have been more likely to react with full brake pedal force more rapidly. No significant difference in assigned following distance was found ( χ2(1) = 0.69, p > 0.05). Similarly, the interaction between preferred and assigned following gap was not significant ( χ2(1) = 0.82, p > 0.05).

Crashes

Next, participant crashes during the final crash event were examined. A crash was defined as the participant vehicle colliding with the immediately preceding vehicle. Participants that drove at the close distance experienced significantly more crashes (M = 0.82) than those who drove at the far distance (M = 0.61) ( χ2(1) = 4.32, p = 0.038) (see figure 3). No difference in collision rate based on preferred following distance was found ( χ2(1) = 0.49, p > 0.05). Similarly, no interaction between preferred and assigned following distance was found ( χ2(1) = 1.27, p > 0.05). Table 8 shows the total number of crashes by preferred and assigned gaps. Because there were an uneven number of participants in each group, table 9 shows the number of participants who did not crash by preferred and assigned following gaps.

Table 8. Total number of crashes by preferred and assigned gaps.

Gap Assigned
Preferred Near Far
Near 20 19
Far 15 11

 

Table 9. Total number of non-crashes by preferred and assigned gaps.

Gap Assigned
Preferred Near Far
Near 7 11
Far 2 8

 

This figure is a bar graph displaying the probability of experiencing a crash during the emergency braking event. The x-axis shows preferred gap, and the y-axis shows probability of crashing and ranges from 0 to 1 in increments of 0.25. The values are grouped by both assigned and preferred following distances (near and far). The mean probability are as follows: assigned near gap and preferred near gap is 0.74, assigned near gap and preferred far gap is 0.88, assigned far gap and preferred near gap is 0.63, and assigned far gap and preferred far gap is 0.58.

Figure 3. Graph. Probability of experiencing a crash based on both preferred and assigned
following time gap.

 

 

 

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101