Truck Platooning Under Real Traffic Conditions: First Insights on Behavioral Adaptations and Gap Preference of Professional Drivers

Abstract

Objective

The aim of the study was to investigate (1) how different gap sizes are perceived by professional truck drivers under real traffic conditions and (2) whether semi-automated platoon driving leads to changes in driving behavior of subsequent manual driving.

Background

Platoon driving is a current branch in the development of automated driving in which two or more vehicles build a convoy. The lead vehicle is controlled manually while following vehicles are electronically coupled to it and drive semi-automated with small gaps in order to achieve a better traffic flow and potential fuel savings.

Method

In a real road experiment, 10 trained professional truck drivers completed a total of 33 test drives with a two-truck platoon on the German highway A9 with a gap size of either 15 or 21 m, in the leading and the following vehicle.

Results

(1) The drivers experienced both gap sizes as comfortable and preferred the smaller gap size of 15 m. (2) Both gap sizes led to significantly higher standard deviation of lane position in post- compared to pre-platoon driving. No significant difference in distance keepings in post- compared to pre-platoon driving occurred. Qualitative data give hints on difficulties, when switching back to regular truck driving.

Conclusion

The results implicate that small gap sizes are perceived as comfortable by drivers and that platoon driving has an influence on subsequent manual driving.

Application

Countermeasures to behavioral adaptations should be considered in order to ensure a safe conduction of platoon driving.

Keywords

truck platoon driving commercial truck driver on-road study human–automation interaction driver behavior vehicle automation

In platoon driving, two or more vehicles build a convoy that is controlled by a manually driven lead vehicle. Succeeding vehicles are connected electronically and follow in short distance. The following vehicles operate semi-automated, which means a system takes over lateral as well as longitudinal control of the vehicle, but the driver still needs to be attentive at all times. This is called Stage 2 automation as defined by the Society of Automotive Engineers (SAE, 2018). Expected advantages are fuel savings, improved traffic flow, and enhanced safety and comfort (Esser & Kurte, 2018; Roland Berger, 2016). Some authors predict fully automated driving by 2030 (Heider et al., 2017; Walker et al., 2001). At present, automated driving is allowed on public roads, under the condition that the driver can still control the vehicle and override or switch off the system if necessary (Vienna Convention 1968, Art. 13; Art. 8; ECE/TRANS/WP.1/145). Therefore, each following truck of a platoon requires a driver, who monitors the system and can intervene at any time. This new monitoring role of the driver, implied by semi-automated driving, can cause fatigue, poorer situation awareness, and increased engagement in secondary tasks (de Winter et al., 2014; Gonçalves et al., 2017). In semi-automated platoon driving, another issue might add up to this. The driver constantly experiences small distances of up to 10 m to the leading vehicle over a prolonged period of time. This not only restricts the field of view and the time frame to react to unexpected events, but can also lead to changes in the distance behavior of subsequent manual driving. There is evidence that drivers of semi-automated following vehicles choose smaller distances after having experienced a period of small gap platoon driving (Brandenburg & Skottke, 2014; Eick & Debus, 2005; Levitan et al., 1998; Skottke et al., 2014). Gouy et al. (2014) has shown that driving next to a truck platoon can lead car drivers to change their driving behavior and keep smaller distances.

It appears that these issues are more pronounced, the smaller the following distance. Following distances are usually measured in time headway (THW), which is defined as “the elapsed time between the front of the lead vehicle passing a point on the roadway and the front of the following vehicle passing the same point” (Evans, 1991, p. 313). Carryover effects on subsequent manual driving were found for THW of 0.3, 0.625, and 1 s (Eick & Debus, 2005; Levitan et al., 1998; Skottke et al., 2014) but not for individually convenient gaps (Eick & Debus, 2005). Participants seem to adapt to small gaps and change their subsequent manual driving behavior in ways of shorter THW and a tendency of impaired lane keeping. Lane keeping quality is usually assessed by calculating the standard deviation of the vehicles’ lane position (SDLP), measured via the distance to the right lane markings.

In terms of gap perception, a first analysis by Yang et al. (2018) suggests that truck drivers prefer longer distances to the lead vehicle. The study was conducted in 2017 with three platoon trucks and nine professional truck drivers within the PATH project in California, USA. Participants could choose time gaps between 0.6 and 1.8 s THW at 55 m/hr. Time gaps of 1.2 and 1.5 s THW were preferred by the drivers. These time gaps are equivalent to an absolute gap size of 26.6 and 33.3 m at a speed of 80 km/hr (Yang et al., 2018). In Europe, real traffic experiments were conducted within the KONVOI project. The trucks were driven by engineers and escorted by the police, and no results on preferred distance were presented (RWTH Aachen Faculty of Mechanical Engineering, 2009). A test drive on public roads within SARTRE project was only exemplary and did not include systematic experimental setup (Jootel, 2012). Therefore, only little is known about platoon gap perception of truck drivers in real traffic.

In contrast to previous platoon trials, this study includes data of test drives with commercial truck drivers and focuses on their experience under real traffic conditions. To our knowledge, no previous tests about gap perception and behavioral adaptations have been conducted with truck platoons in real traffic on European roads before. It was very uncertain how the drivers would react to the small gaps and if they would be willing to trust the platoon system.

From an economic perspective, minimal gaps in platoon driving are desirable because fuel savings increase with decreasing distance between the trucks. Several studies found fuel savings between 8% and 21% in platoon driving with a gap size of 10 m at a velocity of 80 km/hr (Bonnet & Fritz, 2010; Browand et al., 2004; Dávila, 2013; RWTH Aachen Faculty of Mechanical Engineering, 2009; Tsugawa et al., 2011). Taking into account that fuel savings are highest when driving with gaps of up to 5 m, a discrepancy between efficiency and driver comfort emerges. The question of safety and driver comfort needs to be addressed in order to determine the optimal gap size, while considering economic as well as human factor perspectives.

Methods

Participants

Ten professional truck drivers, currently working for a service company of DB Schenker, participated in the project. Drivers of this company were informed about the project via a notice board and by their dispatchers. The selection was based on practical feasibility of the planned test drives in the shift operation. The participation in the project included one-on-one interviews and questionnaires, intensive platoon driving training sessions, participation in experimental drives, and platoon drives within the day-to-day logistic operations. All participants were male, between 29 and 54 years old (M = 39.3, SD = 7.9), and had at least 8 years of experience as professional truck drivers (M = 14.2, SD = 5.6).

Apparatus

The platoon consisted of two trucks plus trailer by the manufacturer MAN. The trucks were fully loaded with dummy weights or real goods throughout the experiment. The trucks were equipped with a platoon driving system developed by MAN. It included automated distance control and automated steering for the following vehicle in active platoon mode. When driving semi-automated in platoon mode, a system detected if participants had their hands on the steering wheel, as this was a legal prerequisite. As another legal requirement, a dead man’s switch was implemented in the following trucks, which demanded a driver reaction every 4 min.

A wearable eyetracking device by Tobii and a mobile 32-channel EEG by Brainproducts was used to record physiological data of the drivers. The results of the physiological data analyses (EEG and eyetracking) will be presented elsewhere.

A short questionnaire was handed out after each test drive. It included the following questions: “Did you trust the platoon system?” (mistrusted: −2 to trusted: 2) and “How did you perceive the gap size?” (very uncomfortable: −2 to very comfortable: 2). After all test drives, another questionnaire, focusing on acceptance of platoon driving and overall gap preference, was completed.

Procedure

Data were collected within the German government funded project Elektronische Deichsel – Digitale Innovation (EDDI) in August 2018 on the German public highway A9 between Munich and Nuremberg. This research complied with the American Psychological Association Code of Ethics and was approved by the Institutional Review Board at Hochschule Fresenius, University of Applied Sciences. Each participant gave his informed consent and received an extensive training on the platoon system. They had theoretical lessons on the functions of the platoon system and learned how it would behave in different situations. The drivers then experienced the system and its functions as driver of the lead and the following vehicle in a driving simulation and on a test site. They trained the interaction with the human–machine interface (HMI) and experienced an emergency braking maneuver. Afterwards they had additional training on public roads at daytime and nighttime. After training was completed, the drivers participated in the experimental drives on the German highway A9. A special permit for test drives was provided within the project requiring a maximum speed of 80 km/hr and a minimum distance of 15 m between the trucks.

Two different gap sizes (15 and 21 m), as well as two driving positions (leader/follower) were tested as independent variables in this study (equivalent to 0.6 s/0.9 s THW at 80 km/hr). The distances were chosen because 15 m was the minimum gap size possible within the special permit. Due to technical prerequisites, the distance could only be changed by steps of 3 m and 22 m is the maximum reasonable distance from an economic perspective, when considering fuel savings (McAuliffe et al., 2017). The order of the experimental conditions (15 m leader, 15 m follower, 21 m leader, 21 m follower) was balanced across subjects using Latin square design.

A total of 40 experimental platoon drives were planned, consisting of a leading vehicle and a following vehicle. Thirteen drives had to be canceled or ended prematurely due to sickness (4), weather conditions (1), technical problems with the platoon system or the research equipment (7), and traffic jam (1). Due to the restricted availability of the drivers and limited time frame for the test drives, only 5 drives were reacquired, which leads to a total number of 33 conducted test drives (Table 1).

Table 1

Amount of Experimental Drives per Condition That Were Conducted

Platoon Position	15 m	21 m	∑
Leader	18 (20)	15 (20)	33 (40)
Follower	18 (20)	15 (20)	33 (40)

Note. The amount of originally planned drives is given in parentheses. The platoon gap size was manipulated to either 15 or 21 m. Participants took test drives in the leading and in the following vehicle

The platoon vehicles started at approx. 06:00 p.m. in Munich and drove to Nuremberg, where the drivers took a half-hour break and changed positions (leader/follower). At approx. 08:00 p.m. they drove the same way back to Munich (Figure 1). One way was considered as one test drive. Five of the experimental drives were conducted as back-and-forth drives without a pause in Nuremberg and without switching positions.

Figure 1

Experimental procedure of the test drives on the German public highway A9.

Each test drive consisted of three sections: pre-platoon manual driving (20 km), platoon driving (82 km), and post-platoon manual driving (20 km). The driver of the rear truck was instructed to follow the leading truck manually during pre- and post-platoon sections. During platoon driving sections the following truck was coupled electronically to the lead truck (gap size of either 15 or 21 m). The trucks were occasionally decoupled due to traffic events (e.g., cut-in vehicles), technical problems, and legal requirements, but were reconnected as soon as possible. The leading truck was also driven manually in pre- and post-platoon sections and with adaptive cruise control (ACC) assistance during platoon sections. Right before and right after each test drive, a short questionnaire about the drivers’ experience in the platoon was completed by both drivers (Figure 1).

Data Analysis

Gap perception ratings were averaged per participant, then mean values were calculated per condition and analyzed with nonparametric Friedman test due to nonnormal distribution.

Qualitative interviews were conducted after completion of the test drives. They were fully transcribed and analyzed by qualitative content analysis.

Driving data were derived from different sensors within the truck. The distance to the vehicle ahead was measured by a radar sensor with a cycle period of 40 ms up to a maximum distance of 255 m. If the sensor could not detect the distance, the distance measurement jumped to a default value of over 6,000 m. In order to adjust for these outliers, default values were replaced by linear regression between the last two data points within the accepted range of 250 m. The velocity of the vehicle was measured every 20 ms in km/hr. A similar linear function was used to smooth the data. Both variables were recomputed onto one timescale and combined to a new variable “THW” afterwards as follows:

T H W [s] = \frac{d i s t a n c e t o v e h i c l e a h e a d [m]}{v e l o c i t y o f e g o v e h i c l e [\frac{k m}{h r}]} \times 3.6

d i s t a n c e t o v e h i c l e a h e a d [m] = \frac{T H W [s]}{3.6} \cdot v e l o c i t y o f e g o v e h i c l e [\frac{k m}{h r}]

Distance to the right lane marking was also detected by the vehicle; this variable was used to compute the SDLP. Identification of the platoon periods was derived from the status of the HMI. The HMI either indicated that the platoon system was inactive (0, 10), that a coupling (40) or decoupling maneuver was conducted (50: driver initiated, 90: system initiated), or that regular platoon mode was active (60). An algorithm identified the time of first platoon coupling and last platoon decoupling. With this information, 15-min pre-platoon and 15-min post-platoon sections were identified. For these sections, mean velocity, mean distance to the vehicle ahead, SDLP, and mean THW were computed with a window length of 1 min. Figure 2 shows exemplary raw data of distance, speed, and HMI status with marked mean values per minute in pre- and post-platoon sections.

Figure 2

Exemplary raw data of a test drive in the following truck. The black dots mark the minute-by-minute mean value starting at first coupling and last decoupling of the platoon. HMI status of 60 identifies active platoon mode. Platoon mode was occasionally interrupted by cut-in vehicles or other traffic events. HMI = human–machine interface.

Driving data were checked for consistency in speed, platoon duration, and ACC usage. A total of 13 drives had to be excluded from the analysis of distance behavior, due to one of the following reasons: (a) speed in pre- or post-platoon sections was less than 70 km/hr due to traffic conditions (traffic jam, temporary construction sites) (2); (b) ACC was activated during pre- or post-platoon driving sections (11). When using ACC, the distance to the vehicle ahead does not reflect actual distance keeping behavior. Participants were therefore asked to switch ACC off during manual driving sections, but did not always comply. Either they did not understand the instruction or they rejected to follow them.

For the analysis of SDLP, three drives in the lead vehicle and four drives in the following vehicle had to be excluded, because speed was less than 70 km/hr during pre- or post-platoon periods. As a result, 30 drives in the lead vehicle and 29 drives in the following vehicle were included in the analysis of SDLP.

For the statistical analysis of driving behavior, mean values were calculated for pre- and post-platoon sections (15 min) and aggregated per participant. Driving data were then analyzed with repeated measures ANOVA.

Results

Perception of Gap Size and Trust Ratings

The drivers rated both gap sizes as comfortable in both driving positions. Overall, 85.2% of all the gap size ratings indicated moderate or high comfort. Nonparametric Friedman test of the ratings revealed no significant differences between gap perception in the different platoon driving conditions (χ² (3) = 6.75, p = .08). Trust in the platoon system was also rated high throughout the test drives. Overall, 90.6% of the ratings indicated moderate or high trust in the current gap size (15 m/21 m). Mean ratings of gap perception and trust are shown in Figure 3.

Figure 3

Mean ratings on the perception of gap size and trust in the platoon systems. Ratings were given on a bipolar scale by the lead and following vehicle drivers after each test drive.

After completion of all test drives, participants were asked for their overall gap size preference in the questionnaire. The following gap choices were presented: <10 m/10 m/10–15 m/15 m/15–20 m/20 m/20–30 m/30 m/>30 m. Four participants preferred a distance of 10 to 15 m between the platoon vehicles, six preferred a distance of 15 m. In the interviews, preferences were explained as follows:

Participant 1: The [short] distance doesn’t disturb me. Because you know that it’s safe. … We mainly drove with 15 m distance, that’s ok. We also tried 21 m but that’s too much … because then, we are too far from each other and cars generally try to enter.

Participant 3: 21 m is more inviting for passenger cars to enter the gap … They rather try to drive into the larger gap. At 15 m it’s less.

Participant 5: There is no difference [between 21 m and 15 m]. You don’t feel the difference … You only feel it if passenger cars try to enter the gap. With 15 m they don’t dare as much, with 21 m they quite do.

Participant 8: 10, 15, 20 m it doesn’t matter, the distance is very short.

Participant 9: The difference between 15 and 20 m isn’t that relevant I would say … but what we identified is: the larger the distance between the trucks, the smaller the threshold of the surrounding traffic to drive into the gap.

Distance Keeping

For the analysis of distance behavior, only data of the following vehicles were analyzed. A within-subject ANOVA with the factor time (pre/post) and gap size (15 m/21 m) was computed using the mean values of 15-min driving sections pre-/post-platoon driving. The main effects of neither time, F(1,11) = 1.384, p = .264, η²= .112, nor gap size, F(1,11) = 0.287, p = .603, η² = .025, were significant. There was no interaction effect, F(1,11) = 0.437, p = .522, η² = .038. To show the course of distance keeping behavior within pre- and post-platoon driving, the sections were divided into 3-min phases. Figure 4 shows the in individual distance behavior of each driver at a gap size of 15 and 21 m, as well as the mean values per phase.

Figure 4

Distance keepings in pre- and post-platoon driving sections. The 15-min pre- and post-platoon sections were divided into 3-min phases. Light gray lines show the individual distance keepings for 15 and 21 m gap size. Dark line shows the mean ± 1 SE of all participants.

In one-on-one interviews, the test drivers provided deeper insight into the subjective perception of distance behavior. Some drivers described a change in distance behavior after they completed the test drives:

Participant 3: I remember, after platoon training, when I was on my way with my normal truck, the first moments it was also like that with the distance. You simply tailgate, this 15-m distance was so normal that you are not afraid. You enter the highway, speed up and then you think, hey, just a moment, we’re not driving platoon! Then you let yourself fall back again, but it always takes a moment.

Participant 7: Actually, I drove platoon for 3 weeks in a row and then normal again. Then you really have to think a moment: This is not a platoon now, you must drive differently now. It doesn’t take long but you have to rethink.

Participant 2: […] if you drive in a platoon, one whole week platoon and then you switch to a normal truck, then you have exactly this, that you drive dive too close to the next truck that you meet somewhere.

Traffic volume could not be controlled but was retraced by the Bavarian Traffic Administration. Minute-by-minute traffic data were available on measuring points near the corresponding manual driving sections on the highway A9 (pre-platoon sections: Allershausen(N) 06:00–07:00 p.m., Greding(S) 09:00–10:00 p.m.; post-platoon sections: Greding(N) 08:00–09:00 p.m., Allershausen(S) 10:00–11:00 p.m.). Data of these measuring points were averaged for 1 hr when the platoons roughly passed them. Traffic intensity was calculated for all 20 drives included in the analysis of distance behavior. In pre-platoon sections, traffic intensity ranged from 14 to 92 vehicles per minute with a median of 70.8. In post-platoon sections, the mean values of vehicles per minute were lower and ranged from 15 to 53, with a median of 19.1.

Lane Keeping

For the analysis of the lane keeping quality, data of the following and the leading vehicles were considered. Drivers of the lead vehicle had to steer during the whole course, drivers of the following vehicle drove with automated steering during platoon sections. They had to keep their hands on the steering wheel at all times. The results showed an increase of SDLP in post-platoon sections for following vehicles only (follower pre: M = .146, leader pre: M = .161, follower post: M = .182, leader post: M = .164). A mixed model ANOVA with time (pre/post) as within-subject factor and driving mode (leader/follower) as between-subject factor was conducted. It showed a significant main effect of time F(1,16) = 13.91, p = .002, $η^{2}$ = .465 and a significant interaction of time and driving mode F(1,16) = 9.40, p = .007, $η^{2}$ = .370. The main effect of driving mode was nonsignificant, F(1,16) = 0.46, p = .834, $η^{2}$ = .003. The 15-min pre- and post-platoon sections were divided into 3-min phases to show the course of SDLP development. Figure 5 shows the in individual lane keeping behavior of each driver in the leading and the following vehicles, as well as the mean values per phase and platoon position.

Figure 5

Standard deviation of lane position (SDLP) during pre- and post-platoon driving sections. The 15-min pre- and post-platoon sections were divided into 3-min phases. Light gray lines show the individual steering behavior in the leading vehicle (dashed) and the following vehicle (solid). Dark lines show the mean ± SE of all drives in the leading vehicle (dashed) and the following vehicle (solid).

In the one-on-one interviews, none of the drivers brought forward any issues with lane keeping after platoon diving. The absolute SDLP in the following vehicle increased by 3.6 cm in post- compared to pre-platoon driving. This difference is very subtle and might therefore not be noticeable for the driver. It is yet unclear if this small increase is safety relevant.

Discussion

In this study, results of the first European on-road experiment of platoon driving with trained professional drivers are presented. Ten drivers participated in several platoon test drives under real traffic conditions with different gap sizes between the platoon vehicles. It was investigated whether semi-automated platoon driving has an effect on subsequent manual driving and how different gap sizes are perceived by the drivers. Results show the acceptance of the small gap sizes as well as significantly altered lane keeping after platoon driving. In the following, the results of distance behavior and gap perception as well as lane keeping are discussed separately.

Distance Behavior and Gap Perception

A discrepancy between convenient and economically reasonable gap sizes was expected. Surprisingly, participants rated both tested gap sizes (15 and 21 m) as convenient and even preferred short gap sizes between 10 and 15 m. In post interviews, they stated that the tested gap sizes felt very similar but triggered different reactions of the surrounding traffic. According to the drivers, passenger cars were less likely to enter into small gaps of 15 m than into the larger gap size of 21 m. This was given as explanation for their preference of smaller gaps. In contrast, Yang et al. (2018) found that drivers preferred gap sizes of 26.6 and 33.3 m (1.2–1.5 s THW at 55 m/hr). This might be due to different highway and traffic prerequisites on the Californian highway between Richmond and Wesley where the study was conducted. On European roads, shorter distances seem to be more familiar and accepted by the drivers. The high reliability of the tested platoon system might have contributed to the results. Participants trusted the system and felt comfortable at whatever distance was configured. The tested drivers knew each other well and drove together for at least 1 week, which might have led to higher trust in the platoon partner and acceptance of short gaps.

There is strong evidence from prior research that short gap platoon driving can affect distance keeping of subsequent manual driving (Eick & Debus, 2005; Levitan et al., 1998; Skottke, 2007). However, in this real road experiment, no effect on distance keeping became apparent. We must be very cautious interpreting these results, as nearly one-third of the drives had to be excluded for this analysis. In contrast to well-controlled simulator studies, confounding variables like the behavior of surrounding vehicles and differing traffic volumes cannot be controlled for in real traffic. High traffic volume leads to an increase of short THW as shown in an empirical analysis of traffic flow on German highways (Geistefeld, 2007). In this study, the traffic volume was high in pre-platoon sections and rather low in post-platoon sections. Therefore, the initial distance keeping value of the test drives might have been underestimated and the effect on subsequent manual driving could be more pronounced in more homogenous traffic situations. Furthermore, the drivers in this study were under constant regard and might have been especially aware of keeping safety distances during the test drives.

Although not statistically significant, the drivers subjectively noticed that they kept shorter distances when switching from platoon driving to regular truck diving. This underlines the need to further investigate if platoon driving can have an effect on distance behavior with other traffic prerequisites or longer platoon driving phases. As one of the main causes of accidents with heavy-duty vehicles, insufficient distance behavior could potentially have big impact (Evers & Auerbach, 2005; International Road Transport Union [IRU], 2007). There are different countermeasures that could proactively prevent the probability of insufficient distance keeping after platoon driving. One approach has already been tested in a simulator study within the KONVOI project. A distance feedback system warned the drivers after decoupling if the distance to the vehicle ahead was smaller than 50 m. With this system, no effects of platoon driving on distance keeping in subsequent manual driving were found, as published in the final report of the project (RWTH Aachen Faculty of Mechanical Engineering, 2009; Wille et al., 2008). The authors concluded that correct distance keeping was due to this feedback system. The impact of continuous THW feedback on distance behavior had also been documented by Fairclough et al. (1997).

Lane Keeping Behavior

Results showed greater SDLP of the following vehicle after platoon driving in comparison to the lead vehicle. Higher SDLP values are associated with sleepiness (Ingre et al., 2006; Krüger et al., 2001) and poor quality of takeover (Merat et al., 2014; Naujoks et al., 2014). The significant interaction effect shows that SDLP was increased only for the following vehicle. Drivers of the lead vehicle drove the same course at the same time and did not show impaired steering. Therefore, we assume that this effect is automation specific. Drivers of the lead vehicle steered manually throughout the whole course, whereas drivers of the following vehicle experienced semi-automated driving and did not have to steer manually. However, due to legal requirements, drivers of the following vehicle had to keep their hands on the steering wheel at all times. They even had to add slight pressure on the steering wheel in order for the hands-off detection to work properly. Reduced SDLP could be explained by ideomotor theory. For an overview see Shin et al. (2010) and Stock and Stock (2004). Action and its effects on the environment are closely connected. Mental representations of the action–effects are built and can trigger the corresponding action if activated endogenously or exogenously. In our case the driver’s action (holding on to the steering wheel) does not correspond to the perceived action–effect (lane keeping of the car). The newly built representation of passive steering might be responsible for the phenomenon of impaired lane keeping after platoon driving that we found. To gain further evidence for this theoretical approach the effect should be analyzed systematically, for example, by simulator studies with forced feedback on the steering wheel. If proven to be true, this might have implications on the design of automated steering systems, that is, provide less or no movement of the steering wheel or take the hands off completely. However, until not further investigated no recommendations can be given on this behalf.

Limitations

As this experiment took place under real traffic conditions on public roads, confounding variables such as traffic volume and weather conditions could not be controlled for. Due to the fact that the participants were commercial truck drivers and unexperienced to test drives and experimental procedure, compliance to the test sequence was another limiting factor (ACC usage during manual driving sections). To still provide acceptable internal validity, a high proportion of data had to be excluded as described above.

Another important point is that one of the drivers of the two-vehicle platoon was always accompanied by a researcher which could have led to more rule-consistent driving behavior.

Another limitation is that only male drivers participated in this study. Female drivers might be less accepting of small gap platoon driving.

Application

The experiment involved a first application and a proof of concept of platoon driving in real traffic in Germany. It was successfully completed without any serious incidents. In contrast to previous assumption the drivers trusted the small gap sizes and stated to have felt comfortable with both gap sizes (15 and 21 m). These findings are expected to be transferable to natural driving situations because of the high external validity of this field study. However, objective data of lane keeping and subjective data of distance keeping showed that behavioral adaptations appeared, which should be further investigated. Technical solutions to further assist drivers after decoupling could be a possible answer. In contrast to our expectation, the comparison of gap size did not show any indication that the smaller gap of 15 m has higher impact on the driver than the larger gap size of 21 m. However, these results should be seen as first insight and confirmed with a larger sample.

Key Points

Small gaps of 15 and 21 m between platoon vehicles were perceived as convenient by professional drivers as tested in an on-road experiment.

The gap size of 15 m was preferred by professional drivers because passenger cars were less likely to enter the gap.

Driving behavior in manual driving was affected by preceding semi-automated platoon driving.

Countermeasures to behavioral adaptations should be considered to ensure a safe implementation of platoon driving.

Footnotes

Acknowledgments

This research was conducted within the research project “Elektronische Deichsel – Digitale Innovation” (electronic drawbar – digital innovation) funded by the German Federal Ministry of Transport and Digital Infrastructure (funding code: 16AVF1031B). We thank the research team of MAN Truck & Bus AG for providing and preprocessing driving data of the platoon trucks. We also thank the Bavarian Highway Administration for providing traffic data of the German highway A9. The results of this research could have an impact on the application of platoon driving in logistic operations of DB Schenker, but they do not implicate financial benefit for any of the authors.

ORCID iD

Sarah-Maria Castritius

Author Biographies

Sarah-Maria Castritius is a human factors researcher at Hochschule Fresenius, Idstein, Hesse, Germany, and a PhD student at Johannes Gutenberg-Universität Mainz. She received her master’s degree in psychology from Julius-Maximilians-University Würzburg, Germany, in 2017.

Christoph Johannes Dietz is a social scientist at Hochschule Fresenius, Idstein, Hesse, Germany. He received his master’s degree in therapy sciences from Hochschule Fresenius Idstein, Germany, in 2016.

Patric Schubert is a data scientist at Hochschule Fresenius, Idstein, Hesse, Germany. He received his PhD from Johann Wolfgang Goethe-University, Frankfurt am Main, Germany, in 2014.

Johanna Moeller is a social scientist at Hochschule Fresenius, Idstein, Hesse, Germany. She received her bachelor’s degree from Hochschule Fresenius Idstein, Germany, in 2017.

Simone Morvilius is a social scientist at Hochschule Fresenius, Idstein, Hesse, Germany. She received her master’s degree from German Sport University Cologne, Germany, in 2015.

Sabine Hammer is a professor for social sciences research at Hochschule Fresenius, Idstein, Hesse, Germany. She received her PhD from Technische Universität Kaiserlautern, Germany, in 2017.

Chung Anh Tran is an associate of Deutsche Bahn AG, Frankfurt am Main, Hesse, Germany. He received his PhD from Karlsruhe Institute of Technology, Germany, in 2011.

Christian T. Haas is a professor for quantitative methods at Hochschule Fresenius, Idstein, Hesse, Germany. He received his habilitation from Goethe-University Frankfurt, Germany, in 2007.

References

Bonnet

Fritz

. (2010). Fuel consumption reduction in a platoon: Experimental results with two electronically coupled trucks at close spacing. In SAE Technical Paper Series (Vol. 1). https://doi.org/10.4271/2000-01-3056.

Brandenburg

Skottke

E.-M

. (2014). Switching from manual to automated driving and reverse: Are drivers behaving more risky after highly automated driving? [Conference session] IEEE 17th International Conference on Intelligent Transportation Systems (ITSC), IEEE. pp. 2978–2983.

Browand

McArthur

Radovich

. (2004). Fuel saving achieved in the field test of two tandem trucks. California PATH Research Report, (6), 147. http://its.berkeley.edu/sites/default/files/publications/UCB/2004/PRR/UCB-ITS-PRR-2004-20.pdf

Dávila

. (2013). SARTRE report on fuel consumption (technical report for European Commission under the seventh framework programme project 233683 deliverable 4.3).

de Winter

J. C. F.

Happee

Martens

M. H.

Stanton

N. A

. (2014). Effects of adaptive cruise control and highly automated driving on workload and situation awareness: A review of the empirical evidence. Transportation Research Part F: Traffic Psychology and Behaviour, 27, 196–217.doi:10.1016/j.trf.2014.06.016

Eick

E.-M.

Debus

. (2005). Adaptation effects in an automated car-following scenario. In Underwood

(Ed.), Traffic and transport psychology. Theory and application (pp. 243–255). Elsevier.

Esser

Kurte

. (2018). Autonomes Fahren- Aktueller Stand, Potentiale und Auswirkungsanalyse, Studie im Auftrag des Deutschen Industrie- und Handelskammertag (DIHK) e: KEConsult. https://doi.org/10.1007/978-3-8348-8619-4_44.

Evans

. (1991). Traffic safety and the driver. Van Nostrand Reinhold.

Evers

Auerbach

. (2005). Verhaltensbezogene Ursachen schwerer Lkw-Unfälle. Mensch und Sicherheit (Vol. M 174). Bundesanstalt für Strassenwesen BASt.

10.

Fairclough

S. H.

May

A. J.

Carter

. (1997). The effect of time headway feedback on following behaviour. Accident Analysis & Prevention, 29, 387–397.doi:10.1016/S0001-4575(97)00005-5

11.

Geistefeld

. (2007). Verkehrsablauf und Verkehrssicherheit auf Autobahnen mit vierstreifigen Richtungsfahrbahnen. [Dissertation]. Ruhr-Universität Bockum. http://www-brs.ub.ruhr-uni-bochum.de/netahtml/HSS/Diss/GeistefeldtJustin/diss.pdf.

12.

Gonçalves

R. C.

Quaresma

Rodrigues

C. M. A

. (2017). Approaches for loss of vigilance in vehicle automation: A meta-analytical study. Proceedings of the Human Factors and Ergonomics Society, 61, 1871–1875.doi:10.1177/1541931213601948

13.

Gouy

Wiedemann

Stevens

Brunett

Reed

. (2014). Driving next to automated vehicle platoons: How do short time headways influence non-platoon drivers’ longitudinal control? Transportation Research Part F: Traffic Psychology and Behaviour, 27, 264–273.doi:10.1016/j.trf.2014.03.003

14.

Heider

H. J.

Schallehn

Schlegel

Stricker

. (2017). An autonomous car roadmap for suppliers. Bain and Company. Munich. http://www.bain.de/en/Images/BAIN_BRIEF_An_Autonomous_Car_Roadmap_for_Suppliers.pdf

15.

Ingre

Akerstedt

Peters

Anund

Kecklund

. (2006). Subjective sleepiness, simulated driving performance and blink duration: Examining individual differences. Journal of Sleep Research, 15, 47–53.doi:10.1111/j.1365-2869.2006.00504.x

16.

International Road Transport Union [IRU] . (2007). ETAC - European truck accident causation. Final report ETAC, European Commission. https://www.iru.org/sites/default/files/2016-01/en-2007-etac-study.pdf.

17.

Jootel

. (2012). Safe road trains for the environment (SARTRE) final report. (technical report for European Commission under the framework 7 programme project 233683). Ricardo UK Limited. http://sartre-project.eu/en/Sidor/default.aspx.

18.

Krüger

H. -P.

Tietze

Hargutt

. (2001). Analyse zur Beurteilung des Verlaufs von Ermüdung [Conference session]: Conference-Proceeding. Regensburg, 43. Tagung experimentell arbeitender Psychologen. Retrieved from http://www2.psychologie.uni-wuerzburg.de/izvw/texte/2001_tietze_hargutt_Zweidimensionale_Analyse.pdf.

19.

Levitan

L. E. E.

Golembiewski

Bloomfield

J. R

. (1998). Human factors issues for automated highway systems. ITS Journal - Intelligent Transportation Systems Journal, 4, 21–47.doi:10.1080/10248079808903735

20.

McAuliffe

B. R.

Croken

Ahmadi-Baloutaki

Raeesi

. (2017). Fuel economy testing of a three vehicle truck platooning system. Laboratory Technical Report, National Research Council Canada, Aerospace Aerodynamics Laboratory, No. LTR-AL-2017- 0008. https://doi.org/10.4224/23001922.

21.

Merat

Jamson

A. H.

Lai

F. C. H.

Daly

Carsten

O. M. J

. (2014). Transition to manual: Driver behaviour when resuming control from a highly automated vehicle. Transportation Research Part F: Traffic Psychology and Behaviour, 27, 274–282.doi:10.1016/j.trf.2014.09.005

22.

Naujoks

Mai

Neukum

. (2014). The effect of urgency of take-over requests during highly automated driving under distraction conditions. In Proceedings of the 5th International Conference on Applied Human Factors and Ergonomics AHFE 2014 (pp. 2099–2106). AHFE Conference.

23.

Roland Berger . (2016). Automated Trucks - The next big disruptor in the automotive industry? Roland Berger GmbH. https://www.rolandberger.com/publications/publication_pdf/roland_berger_automated_trucks_20160517.pdf

24.

RWTH Aachen Faculty of Mechanical Engineering . (2009). Verbundprojekt KONVOI : Entwicklung und Untersuchung des Einsatzes von elektronisch gekoppelten Lkw-KONVOIs. (Abschlussbericht). https://www.tib.eu/de/suchen/download/?tx_tibsearch_search%5Bdocid%5D=TIBKAT%3A629521301&cHash=a554f0bbc091ccf5d4d51869f93fdcfc#download-mark.

25.

SAE . (2018). Taxonomy and definitions for terms related to on-road motor vehicle automated driving systems. SAE International, J3016_2018, 1–12. https://www.techstreet.com/standards/sae-j3016_201806?product_id=2016943

26.

Shin

Y. K.

Proctor

R. W.

Capaldi

E. J

. (2010). A review of contemporary ideomotor theory. Psychological Bulletin, 136, 943–974.doi:10.1037/a0020541

27.

Skottke

E. -M.

Debus

Wang

Huestegge

. (2014). Carryover effects of highly automated convoy driving on subsequent manual driving performance. Human Factors: The Journal of the Human Factors and Ergonomics Society, 56, 1272–1283.doi:10.1177/0018720814524594

28.

Skottke

E.-M

. (2007). Automatisierter kolonnenverkehr und adaptiertes fahrverhalten. Untersuchungen des abstandsverhaltens zur bewertung möglicher künftiger verkehrsszenarien.

29.

Stock

. (2004). A short history of ideo-motor action. Psychological Research, 68, 176–188.doi:10.1007/s00426-003-0154-5

30.

Tsugawa

Kato

Aoki

.pp. (2011). An automated truck platoon for energy saving. In IEEE International Conference on Intelligent Robots and Systems (pp. 4109–4114). IEEE. https://doi.org/10.1109/IROS.2011.6048157.

31.

Walker

G. H.

Stanton

N. A.

Young

M. S

. (2001). Where is computing driving cars? International Journal of Human-Computer Interaction, 13, 203–229.doi:10.1207/S15327590IJHC1302_7

32.

Wille

Röwenstrunk

Debus

. (2008). KONVOI: Electronically coupled truck convoys. In de

Waard

F. O.

Flemisch

Oberheid

Brookhuis

K. A.

(Eds.), Human factors for assistance and automation (pp. 243–256). Shaker Publishing.

33.

Yang

Shladover

S. E.

X.-Y.

Spring

Nelson

Ramezani

. (2018). A first investigation of truck drivers’ on-the-road experience using cooperative adaptive cruise control. UC Berkeley. https://escholarship.org/uc/item/92359572.