A Physiological Evaluation of Driver Workload in the Lead Vehicle of an Autonomous Truck Platoon Using Bio-Signal Analysis

Abstract

The evaluation of driver workload in the lead vehicle of a driver-following autonomous truck platoon was conducted using bio-signal analysis. In this study, a single driver operated the lead vehicle while the second and third trucks followed autonomously. Three professional truck drivers (38 ± 4 years old, male) participated in the experiment. During driving, wearable sensors measured heart-rate variability indices, body acceleration, and skin temperature. The heart rate and body acceleration were sampled at 128 Hz (7.8 ms intervals), while skin temperature was recorded at 1 Hz. Each participant underwent three measurement sessions on different days, with each session lasting approximately 30–40 min. Statistical analysis was performed using repeated-measures ANOVA to determine significant differences across conditions and days. The results indicated that compared to solo driving, driving the lead vehicle of the autonomous platoon significantly increased skin temperature (p < 0.001), suggesting a higher physiological workload. This study provides insight into the physiological impact of autonomous platooning on lead-vehicle drivers, which is crucial for developing strategies to mitigate driver workload in such systems.

1. Introduction

In recent years, with the progression of climate change, reducing greenhouse gas (GHG) emissions has become a global priority. In particular, reducing GHG emissions in the transportation sector is recognized as a key area of focus due to its significant contribution to overall emissions. In this context, truck platooning has garnered attention as an advanced technology aimed at lowering fuel consumption and minimizing environmental impact [1]. Truck platooning enables multiple vehicles to travel in close formation, primarily utilizing vehicle-to-vehicle (V2V) communication and automated driving technology. This system allows trailing vehicles to reduce the air resistance generated by the lead vehicle, thereby improving overall fuel efficiency. Additionally, interactions with highway infrastructure play a crucial role in enhancing the environmental benefits of platooning. It has been suggested that integrating platooning with smart highways could further reduce GHG emissions compared to conventional highways, as it optimizes traffic flow and minimizes unnecessary acceleration and deceleration [2,3,4,5,6,7,8,9,10,11,12,13].

Beyond fuel savings, truck platooning is expected to enhance cost efficiency in the logistics industry. However, several social factors must be addressed, including the necessary infrastructure development and changes in truck drivers’ attitudes toward this technology. Research is also being conducted on drivers’ behavioral adaptation and their preferences regarding inter-vehicle distances from a safety perspective. While demonstration experiments with professional drivers indicate a relatively high level of acceptance regarding platooning safety, concerns have been raised about visibility and responses to sudden braking [14,15,16,17,18,19,20,21]. Education and training are essential for overcoming psychological barriers among drivers and promoting the widespread adoption of this technology.

Furthermore, government policy support and regulatory development are critical to facilitating its large-scale implementation. While safety remains the top priority, little research has been conducted on predicting drivers’ conditions based on their physiological signals, largely because platooning is still a relatively new technology. To address this gap, expanding safe demonstration experiments is necessary. In this study, we aim to maximize the potential of platooning technology and contribute to the development of a sustainable transportation system. Specifically, we will analyze the electrocardiogram, bio-acceleration, and body surface temperature of the driver leading the truck platoon, comparing these metrics to solo-driving conditions to identify key physiological differences. The specific objective of this research is to clarify the physiological stress on the driver operating the lead vehicle in a platoon of autonomous trucks. Previous research has often focused on the behavior and safety of autonomously driven following vehicles, and there have been limited empirical data on the psychological and physiological stress placed on the lead-vehicle driver. This research quantitatively evaluates the difference in stress compared to normal driving using bio-signals such as heart-rate variability and skin temperature, and is novel in that it highlights stress factors specific to the lead-vehicle driver.

2. Materials and Methods

2.1. An Experiment of Truck Platooning Technology with Unmanned Vehicles on the Trailing Edge on Expressways

In this study, a platoon of three heavy-duty trucks was used, where only the lead vehicle was manually driven by the participant, and the second and third trucks followed autonomously; we investigated the physiological responses of drivers of the lead vehicle in a truck platoon. This experiment was conducted as part of a broader initiative aimed at addressing issues such as the shortage of truck drivers, the aging of the workforce, and improving fuel efficiency.

The Ministry of Land, Infrastructure, Transport and Tourism (MLIT) and the Ministry of Economy, Trade and Industry (METI) have been developing truck platooning technology, including automated following vehicles, and analyzed the anonymized data that could not be linked when conducting a demonstration experiment on the New Tomei Expressway (https://www.mlit.go.jp/report/press/jidosha07_hh_000362.html accessed on 16 April 2025).

In this experiment, three trucks formed a convoy, with the focus on the driver of the lead vehicle, who was driving at a speed of 80 km/h with a distance of approximately 9 m between vehicles. The lead vehicle was driven manually, but the following vehicles were equipped with an automatic driving system to maintain the convoy. Physiological data such as heart rate were recorded for the three drivers of the lead vehicle. The experiment was conducted on a real driving route, including expressways, with each session lasting 30–40 min and repeated on three different days. Measurements were taken during daytime under clear weather conditions, with cabin temperature maintained using air conditioning. During driving, physiological signals (heart-rate variability, body acceleration, and skin temperature) were recorded in real time using wearable sensors. All driving sessions followed the same route to ensure consistency.

2.2. Dataset

In this study, physiological responses were evaluated by analyzing heart-rate variability (HRV) indices, bio-acceleration, and skin temperature measured in the experiment. Data were recorded using wearable sensors, and the HRV index was derived from the RR intervals of the electrocardiogram (ECG) signal. Bio-acceleration and skin temperature were also monitored continuously at the same time. Data were collected at various stages of the driver’s operation. Wearable sensors (myBeat WHS-2, Union Tool Co., Tokyo, Japan) were used to measure RR intervals for HRV analysis and body acceleration at a sampling rate of 128 Hz, and skin temperature was recorded at 1 Hz using an external sensor. All measurements were synchronized and recorded continuously during the driving tasks. The conditions and schedule for each participant were counterbalanced to reduce order effects. Data were segmented by driving condition (solo vs. platoon). Experimental errors were minimized by repeating sessions on different days, using professional drivers, and maintaining consistent environmental conditions. Noise was reduced through sensor calibration and applying preprocessing filters to the bio-signal data.

2.2.1. Measurement Protocol

To evaluate the driving workload of the lead-vehicle driver in a truck platooning system with autonomous following vehicles using physiological signal analysis, an experiment was conducted with three truck drivers (mean age: 38 ± 4 years, all male). In this study, only the lead vehicle had a human driver, while the second and third trucks operated autonomously without human intervention.

Each measurement session lasted approximately 40 min, with three drivers participating once per day (Figure 1). In the first session, Subject A drove the lead vehicle (Frame 1). Since the trucks were equipped with an autonomous driving system, the driver was present only as a precaution and did not actively control the vehicle. In the second and third sessions, Subject A was seated in one of the following trucks as a precautionary measure (Frames 2 and 3), but again, no manual driving was performed due to the autonomous system. After each session, the driver returned to the parking lot in solo-drive mode (Solo).

Subject B took the lead-vehicle position in the second session, while Subject C took the lead-vehicle position in the third session. This experiment was conducted over three days. On the second day, Subject A drove the lead vehicle in the second session, and on the third day, Subject A drove the lead vehicle in the third session.

A counterbalanced design was used to mitigate order effects by varying the sequence of trials for each participant. This study employed a fully counterbalanced design to ensure equal distribution of sequence effects across all participants.

2.2.2. Physiological Data

Physiological data were measured using the myBeat system from Union Tool Co., Ltd. (Tokyo, Japan) (https://www.uniontool-mybeat.com/ accessed on 16 April 2025). To minimize noise contamination, R-R intervals (RRI) were measured using the NASA lead method, which reduces the influence of electromyography artifacts by recording the potential difference between the upper and lower sternum. Noise processing was conducted using the RRI Analyzer software (Version 2, Union Tool Co., Tokyo, Japan), which allows users to specify the analysis timing and data range. Therefore, noisy data from the first 30 s after the start of measurement were excluded from the analysis. The recorded data were processed using RRI Analyzer, which converted the raw data into CSV format for further analysis. ECG signals were acquired using myBeat Disposable Electrodes UIR, which were specifically designed for myBeat devices to ensure stable signal acquisition. The sampling frequency for ECG RR intervals and bio-acceleration was set at 128 Hz, with a data update cycle of 7.8 ms. Skin surface temperature was recorded at a sampling frequency of 1 Hz, with a data update cycle of 1 s. These data-acquisition settings ensured high temporal resolution for analyzing physiological responses during truck platooning.

The data measured by the heart-rate sensor in the chest was automatically converted into a heart-rate variability index by the RRI Analyzer (Union Tool Company product, https://www.uniontool-mybeat.com/SHOP/8600025.html accessed on 16 April 2025). This produced the following outputs: RRI (heart-rate interval), low-frequency component (LF), high-frequency component (HF), LF/HF ratio, and VLF (very-low-frequency component). In addition, the breathing frequency (Hz) was calculated from the HF component. The classification of heart-rate variability frequency components was based on the ESC Task Force (https://www.escardio.org/The-ESC/Governance/ESC-Task-Forces accessed on 16 April 2025).

The method for calculating the respiratory frequency (Hz) from the HF component is as follows. First, we performed RR interval series interpolation and equidistant resampling. Since equidistant data are required for HRV analysis, resampling is performed. In this study, cubic spline interpolation was performed at 2 Hz. Next, we detected the peak frequency contained in the high frequency (HF) band (0.15 to 0.40 Hz) for a power spectrum analysis of the HF band. There are various methods for spectral analysis, including FFT (Fast Fourier Transform), Lomb–Scargle Periodic Analysis, which can be used to analyze non-equidistant data, and AR (Autoregressive) Model Analysis, but in this study, the general FFT method was used. The extraction of the respiratory frequency was obtained by taking the peak frequency (position of the maximum value) of the HF-component power spectrum as the respiratory frequency (Hz).

If the peak of the power spectrum is 0.25 Hz, the following is obtained.

Respiratory cycle = 1/0.25 = 4 s

Respiratory rate = 0.25 × 60 = 15 times/minute

In this measurement, the respiration curve was calculated using the above formula because direct measurement using a respiration band or thermistor sensor was not performed.

2.3. Statistical Analysis

Statistical analysis was performed using repeated-measures analysis of variance (ANOVA) to examine significant differences in physiological responses across different driving conditions. The dependent variables included heart-rate variability (HRV) indices, biological acceleration, and skin temperature. Repeated-measures ANOVA was chosen to account for within-subject variability, as the same drivers were measured under different conditions.

Prior to conducting ANOVA, data were tested for normality using the Shapiro–Wilk test. If the assumption of normality was violated, appropriate data transformations were applied. Additionally, Mauchly’s test was performed to assess the sphericity assumption, and if sphericity was violated, the Greenhouse–Geisser correction was applied to adjust the degrees of freedom.

Post hoc pairwise comparisons were conducted using Bonferroni-corrected t-tests to control for multiple comparisons. All statistical analyses were performed using SAS 9.4 (SAS Institute Japan Ltd., Tokyo, Japan), with the significance level set at p < 0.05.

2.4. Ethics Review

This dataset consists of de-identified, non-linkable data, with all personally identifiable information, such as names, removed from the recorded physiological data. To ensure privacy, random IDs or numbers were assigned to the data, preventing the identification of specific individuals. Therefore, from the perspective of Japan’s Personal Information Protection Law, this dataset does not constitute personal information.

However, since this research involves the analysis of human physiological data, it was conducted within the scope of an analysis method previously approved by the Nagoya City University Hospital Ethics Review Committee (Approval No.: 60-18-0211, “Analysis of Time-Series Physiological Signals Acquired by Wearable Sensors”, Approval Date: 22 March 2019). Additionally, the experiment was conducted in accordance with the Declaration of Helsinki, which outlines ethical principles for medical research involving human subjects. Informed consent was obtained from all participants, who voluntarily participated after receiving a full explanation of the study.

3. Results

The results, analyzed using repeated-measures ANOVA, indicate no significant differences in HRV indices and biological acceleration among the three drivers operating the lead vehicle in truck platooning (

Table 1. Changes in biomarkers with driving condition (individual participants).

	F1	F2	F3	Rest *	Solo
Driver A
HR, bpm	89 ± 5	87 ± 4	88 ± 3	94 ± 7	88 ± 5
VLF, ms	43 ± 11	51 ± 15	40 ± 13	55 ± 17	38 ± 14
LF, ms	35 ± 13	32 ± 12	25 ± 13	72 ± 13	23 ± 9
HF, ms	23 ± 8	22 ± 9	15 ± 10	35 ± 15	16 ± 9
LF/HF	2.5 ± 1.2	2.6 ± 1.2	3.4 ± 1.8	5.0 ± 1.7	2.6 ± 1.5
Fb, Hz	0.21 ± 0.09	0.21 ± 0.09	0.25 ± 0.10	0.2 ± 0.03	0.27 ± 0.10
BM, mG	0.52 ± 0.35	0.56 ± 0.39	0.53 ± 0.31	1.03 ± 0.52	0.55 ± 0.57
Skin temp, °C	30.4 ± 0.8	29.7 ± 0.8	28.5 ± 0.7	28.7 ± 0.1	26.9 ± 3.7
Driver B
HR, bpm	106 ± 5	105 ± 5	104 ± 4	118 ± 5	97 ± 6
VLF, ms	17 ± 6	18 ± 9	17 ± 7	22 ± 7	23 ± 9
LF, ms	13 ± 5	17 ± 7	16 ± 8	11 ± 6	21 ± 6
HF, ms	6 ± 2	7 ± 3	9 ± 7	5 ± 2	11 ± 3
LF/HF	6.3 ± 4.2	7.0 ± 5.4	4.3 ± 2.8	6.1 ± 6.3	4.0 ± 2.1
Fb, Hz	0.23 ± 0.09	0.29 ± 0.10	0.28 ± 0.09	0.27 ± 0.08	0.26 ± 0.08
BM, mG	0.54 ± 0.29	0.50 ± 0.28	0.50 ± 0.32	1.75 ± 1.65	0.53 ± 0.41
Skin temp, °C	30.8 ± 0.5	29.8 ± 0.2	29.5 ± 0.3	30.3 ± 0.4	28.7 ± 0.3
Driver C
HR, bpm	115 ± 10	96 ± 8	101 ± 7	106 ± 7	82 ± 4
VLF, ms	24 ± 9	47 ± 17	29 ± 13	31 ± 16	34 ± 13
LF, ms	16 ± 8	45 ± 25	22 ± 7	34 ± 17	30 ± 12
HF, ms	8 ± 3	24 ± 14	10 ± 5	16 ± 6	16 ± 7
LF/HF	4.8 ± 2.4	4.4 ± 2.5	5.7 ± 3.3	5.4 ± 3.5	4.2 ± 2.9
Fb, Hz	0.18 ± 0.04	0.17 ± 0.03	0.16 ± 0.01	0.17 ± 0.02	0.17 ± 0.05
BM, mG	0.69 ± 0.67	0.54 ± 0.3	0.73 ± 0.85	1.08 ± 0.58	0.72 ± 0.76
Skin temp, °C	31.7 ± 0.7	29.9 ± 0.4	29.2 ± 0.4	30.2 ± 0.3	29.0 ± 0.3

* Data are mean ± SD. All participants smoked during the resting period. HR = heart rate; VLF = very-low-frequency component; LF = low-frequency component; HF = high-frequency component; LF/HF = LF-to-HF power ratio; Fb = estimated breathing frequency; BM = body movement; Skin temp = skin temperature.

,

Table 2. Effects of driving condition on biomarkers (repeated-measures ANOVA of three drivers).

	F1	F2	F3	Solo	F (3, 443)	p *
HR, bpm	103 ± 4	96 ± 4	98 ± 4	89 ± 4	79.14	<0.0001
VLF, ms	28 ± 7	38 ± 7	29 ± 7	32 ± 7	13.22	<0.0001
LF, ms	21 ± 4	31 ± 4	21 ± 4	25 ± 4	13.46	<0.0001
HF, ms	12 ± 3	17 ± 3	11 ± 3	15 ± 3	10.06	<0.0001
LF/HF	4.5 ± 0.8	4.7 ± 0.8	4.5 ± 0.7	3.6 ± 0.7	4.46	0.004
Fb, Hz	0.21 ± 0.03	0.23 ± 0.03	0.23 ± 0.03	0.24 ± 0.03	2.37	0.07
BM, mG	0.58 ± 0.07	0.53 ± 0.07	0.58 ± 0.07	0.6 ± 0.06	0.34	0.7
Skin temp, °C	31 ± 0.4	29.8 ± 0.4	29.1 ± 0.4	28.2 ± 0.4	78.54	<0.0001

* Values are the least-square means ± standard error of the means. The significance of the effect of driving condition is indicated. Statistically significant differences are shown (p < 0.05).

and

Table 3. Differences in biomarkers between driving conditions (post hoc multiple comparisons for three drivers).

	F1-F2	F1-F3	F1-S	F2-F3	F2-S	F3-S
HR	<0.0001	<0.0001	<0.0001	1	<0.0001	<0.0001
VLF	<0.0001	1	0.09	<0.0001	0.0003	0.2
LF	<0.0001	1	0.1	<0.0001	0.0005	0.05
HF	0.0002	1	0.1	<0.0001	0.03	0.01
LF/HF	1	1	0.07	1	0.01	0.08
Fb	0.8	0.2	0.05	1	1	1
BM	1	1	1	1	1	1
Skin temp	<0.0001	<0.0001	<0.0001	0.005	<0.0001	<0.0001

Values are adjusted for Type I error using the Bonferroni correction method, which controls the family-wise error rate by dividing the significance level (α) by the number of comparisons. This adjustment reduces the likelihood of false positives when performing multiple statistical tests.

, Figure 2). However, body temperature was significantly higher during lead-vehicle operation (p < 0.001).

4. Discussion

These findings suggest that physiological responses, particularly skin temperature, may be influenced by the demands of lead-vehicle operation in truck platooning. The significant increase in skin temperature during lead-vehicle operation suggests a potential physiological response to driving demands in truck platooning. Skin temperature is influenced by autonomic nervous system activity, particularly by sympathetic and parasympathetic balance [22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37]. While acute stress is typically associated with peripheral vasoconstriction and decreased skin temperature due to sympathetic activation, prolonged cognitive load or sustained attention can lead to increased skin temperature due to parasympathetic rebound and enhanced metabolic activity [38,39,40].

In this study, the absence of significant differences in heart-rate variability (HRV) indices and biological acceleration implies that conventional autonomic nervous system markers did not capture substantial changes in physiological stress levels. However, the increase in skin temperature may indicate a subtle stress response or heightened cognitive workload during lead-vehicle operation. Previous research has reported similar findings in tasks requiring sustained attention and decision making, where increased skin temperature was associated with cognitive and emotional engagement rather than acute stress responses. An observed increase in skin temperature under platoon driving aligns with prior findings on stress-related thermoregulatory responses. These results highlight the importance of considering multiple physiological indicators when assessing driver workload and stress in truck platooning. Future studies should explore the relationship between skin temperature and other biomarkers of stress, such as cortisol levels or electrodermal activity, to gain a more comprehensive understanding of driver physiological responses in automated driving environments [41].

This study has several limitations that should be acknowledged. The main limitations include the small sample size (three participants), limited diversity (all male, professional drivers), and the short duration of each session. First, the small sample size of only three drivers limits the generalizability of the findings. Individual differences in physiological responses, such as autonomic regulation and stress sensitivity, may have influenced the results, making it difficult to draw broad conclusions about drivers operating the lead vehicle in truck platooning. Future studies should include a larger sample size to improve statistical power and ensure the robustness of the findings. Second, while skin temperature showed a significant increase, other physiological indicators such as heart-rate variability (HRV) and biological acceleration did not exhibit significant changes. This suggests that skin temperature alone may not fully capture the complexity of stress responses during lead-vehicle operation. Additional physiological measures, such as cortisol levels, electrodermal activity, or subjective stress assessments, could provide a more comprehensive understanding of driver workload and stress. Third, the study did not account for potential confounding factors such as ambient temperature, individual differences in thermoregulation, or prior fatigue levels. These factors could have influenced skin-temperature variations and should be controlled in future research. While the study focuses on physiological workload, it does not include subjective measures of cognitive load (e.g., NASA-TLX questionnaire) or behavioral metrics (e.g., steering-wheel movements, brake/accelerator usage frequency).

Despite these limitations, this study provides valuable insights into the physiological responses of drivers in truck platooning and highlights the need for further research with a more extensive dataset and additional stress biomarkers. Although a fully autonomous driving society is still a long way off, truck platooning is expected to play a crucial role in ensuring efficient logistics until that vision becomes a reality. Given the aging truck-driver population and ongoing labor shortages, platooning can help optimize transportation efficiency and alleviate some of the burdens on the logistics industry.

However, the successful implementation of truck platooning requires not only technological advancements but also a thorough understanding of the physiological and cognitive demands placed on drivers, particularly those operating the lead vehicle. Ensuring the safety and well-being of these drivers is essential for maintaining smooth and reliable platooning operations. This study contributes to this goal by providing insights into the physiological responses of lead-vehicle drivers, highlighting the need for careful monitoring of their workload and stress levels. By identifying key physiological indicators that may impact driver performance, this research plays a part in enhancing the safety and sustainability of truck platooning, ultimately supporting the transition toward more automated and efficient transportation systems. It can be applied to designing workload-management strategies for lead drivers in autonomous truck platoons, such as adaptive rest scheduling or driver support systems, to enhance safety and reduce fatigue in long-haul transport operations.

5. Conclusions

This study analyzed the physiological signals of lead-vehicle drivers in truck platooning and found a significant increase in skin temperature during operation (p < 0.001). This suggests a potential physiological response to cognitive workload and stress, even though no significant changes were observed in heart-rate variability or biological acceleration.

Given the growing labor shortage and aging workforce in the trucking industry, truck platooning is expected to contribute to more efficient logistics. Ensuring the safety and well-being of lead-vehicle drivers is crucial for its successful implementation.

Future research should address the limitations of this study, including the small sample size, and incorporate additional physiological and psychological stress markers to gain a more comprehensive understanding of driver workload.