Understanding the Workload of Remote Truck Operators with Discrete Event Simulation

Abstract

This study employs a discrete event simulation (DES) model to understand the dynamic workload of remote truck operators managing partially-automated trucks. The DES model uses operator queues and event generators simulating automated truck events and leverages data from the California DMV’s disengagement database and driving simulation experiments. Disengagement data were partitioned into three groups by disengagement frequency: low, moderate, and high and separate arrival time distributions were developed for each group. Simulations from the model suggest that for companies with low disengagement rates, operator utilization will likely remain below minimal thresholds to prevent boredom. In contrast, companies with moderate or high disengagement rates both exceed operator utilization capacity and generate prolonged wait times as more trucks are controlled. These findings suggest that calibrating remote truck control to human capabilities will be challenging. A sensitivity analysis suggests that accurately estimating disengagement rates will be crucial for model accuracy and predictive performance.

In 2022, the Federal Motor Carrier Safety Administration (FMCSA) reported 168,738 truck crashes in the United States (Federal Motor Carrier Safety Administration, 2024), underscoring the pressing need for enhanced road and truck driver safety measures. Concurrently, data from the American Trucking Association (ATA) indicated that there were 3.54 million truck drivers employed in the U.S. in the same year (American Trucking Associations, 2024), alongside a significant shortage of approximately 78,000 drivers (American Trucking Associations, Inc., 2022). This highlights a critical need for solutions to safely meet the increasing demand for truck drivers. One proposed solution to address this challenge is the deployment of automated vehicle (AV) technology, specifically highly automated trucks that require operator intervention only outside their operational design domain. In such scenarios, remote operators could supervise and periodically assist these vehicles. To effectively address the driver shortage, it is anticipated that these remote operators would oversee multiple trucks during their shifts.

While substantial research has been conducted on the technical challenges associated with remote operation, such as latency (Neumeier et al., 2019) and control (Kang et al., 2018) issues, there has been minimal focus on the impact of this new paradigm on the operators themselves. The concept of remote trucking supervision closely mirrors the supervisory control of multiple unmanned vehicles (UVs), where discrete event simulation (DES) has been extensively used to model supervisory control within controlled research environments (Donmez et al., 2010). This study aims to bridge the gap between the theoretical aspects of supervisory control and its practical application in commercial trucking. By developing a DES model using real disengagement data from the California Department of Motor Vehicles (CDMV) autonomous vehicle disengagement reports (State of California Department of Motor Vehicles, 2024) and data from a driving simulation study, this research seeks to answer two critical questions: how operator utilization (the estimated ratio of operator remote driving time to full working time) and truck wait time (the estimated time from disengagement to the start of teleoperation) change with the number of trucks, and how sensitive these metrics are to errors in the estimated time between disengagement and teleoperation time.

The DES model required the specification of operator queues (human drivers) and event generators (automated trucks). Each event generator was defined by two distributions: one for interarrival times (the time between each event) and one for service times (the duration needed to manage an event). Interarrival time distributions were estimated from the CDMV disengagement database using exponential distributions, while service time distributions were derived from a driving simulation study where operators assisted vehicles in navigating around obstacles before resuming control, modeled using normal distributions (Xiao et al., 2022). To fit the interarrival time distributions, the mileage

between each disengagement event reported in the CDMV database was converted into the expected time between disengagements, and companies were clustered into three groups based on their mean disengagement frequency.

The model’s effectiveness was evaluated using two outputs: operator utilization (the ratio of time spent working on trucks to overall simulation time) and average truck wait times (the duration trucks wait in the operator’s queue before being managed). Optimal workloads typically fall between 20% and 80% operator utilization (Nehme, 2009; Yerkes & Dodson, 1908), with minimal wait times to prevent traffic disruptions. A sensitivity analysis was conducted across interarrival times, service time means, and standard deviations, estimating operator utilization and wait times under varying error assumptions (10%–200%).

Findings indicated that for the low disengagement frequency group, operator utilization remained below 20% even with up to 20 trucks, suggesting potential underutilization and risk of operator boredom. Conversely, operators in the high and medium disengagement frequency groups reached 20% utilization with just two and four trucks, respectively, and surpassed 80% utilization with 8 and 17 trucks, respectively. Average truck wait times showed significant increases in the high and medium groups, reaching a 1-min threshold at 8 and 16 trucks, respectively. These results suggest that developing remote truck technology that keeps operators sufficiently engaged without causing traffic disruptions due to waiting trucks will be challenging.

The sensitivity analysis revealed that utilization and average wait time estimates were most sensitive to arrival intervals, followed by service time means and standard deviations. This indicates that accurate estimates of disengagement rates are crucial for understanding operator workloads and truck wait times in real-world operations. The study highlights the complexities of implementing remote supervision for highly automated trucks, particularly in balancing operator workload and minimizing truck wait times. These findings underscore the importance of precise disengagement rate estimates and suggest that while remote supervision could help address the driver shortage, careful consideration of operator engagement and truck management is essential to ensure road safety and operational efficiency.