Supply chain and logistics management in the emerging age of autonomous trucking

Ground-based autonomous vehicle (AV) technology has been rapidly evolving in recent years. Broad-scale implementations of autonomous vehicles are expected to bring several benefits such as reduced drivers’ stress, decreased vehicle accidents, and reduced fuel consumption. While much of the AV attention thus far has concentrated in passenger transportation, the consensus among industry experts is that the promise of AV technology will be sooner realized in the freight transport setting due to its less complex driving environment.

Indeed, the invention of AV technology is increasingly perceived to potentially mark an epoch in the history of freight transportation, making the potential impact of autonomous vehicles particularly relevant to supply chain and logistics (SC&L) organizations. Among various settings for AV applications—such as closed venues like those in agriculture and mining, and yard operations like those in port terminals, rail yards, and warehouses—the transformative magnitude is notable for road transportation due to its dominant role in freight movements.

Certainly, the widespread deployment of autonomous trucks (AT) in road freight transportation has yet to come, but when autonomous trucking becomes mainstreamed, its implications for SC&L organizations will extend far beyond the freight transport operations themselves. Not only is transportation the largest cost component in business logistics expenditure, but it also critical to demand fulfillment and has significant influence on supply chain network design.

Given the rapidly evolving nature of AT technologies and their compelling potentials in road freight transportation, this article draws insights from research conducted at Penn State Center for Supply Chain Research. It discusses potential deployment scenarios and explores, through the lens of the logistics triad, possible changes in SC&L management as a result of its broad-scale deployments.

Getting the lay of the autonomous-truck land

To provide an essential understanding of AT technologies and set the scene for subsequent discussions, AT technology and the current state of AT development are highlighted as follows.

Levels of vehicle automation: Automated vehicles versus autonomous vehicles

The progression in AV technology is echoed in the widely accepted taxonomy of vehicle automation developed by the Society of Automotive Engineers International (SAE). Figure 1 depicts SAE’s five levels of vehicle automation—with human-driving engagement ranging from “all time” at Levels 0-2, “certain time or certain driving environment” at Levels 3-4, to “none” at Level 5. In the United States, vehicles with Levels 3-5 automated systems are termed “highly automated vehicles” and are focused scopes of the U.S. Department of Transportation’s Automated Vehicles Comprehensive Plan released in January 2021.

 

Connected vehicles, connected autonomous vehicles, and truck platooning

Another vehicle technology concept that is frequently discussed together with AV technology is connected vehicle (CV) technology that enables bidirectional communication between vehicles, and between vehicles and infrastructure. While the two technological categories may be implemented separately and connectivity is not a required feature of autonomous vehicles, the synergistic effects between them give rise to the development of connected autonomous vehicles (CAVs), which incorporate both AV and CV technologies (see Figure 2).

 

An application of CAV technology, truck platooning systems leverage different levels of SAE’s vehicle automation and can be distinguished into three generations as follows:

• First-generation truck platooning: Fully manned platoon (at least SAE’s Level 3 automation required). Drivers are behind the wheels of both leader and follower trucks forming the platoon. The leading-truck driver is the main agent and is constantly in control, while drivers in following trucks either rest, perform driving operations, or monitor autonomous driving mode.
• Second-generation truck platooning: Hybrid platoon (at least SAE’s Level 4 automation required). Only the leader truck is manned, while the following trucks operate autonomously without a human driver onboard. 
• Third-generation truck platooning: Driverless platoon (SAE’s Level 5 automation required). There is no driver behind the wheel of any truck forming the platoon. All trucks operate autonomously without a human driver onboard.

The state of the art: En route to Level 4 automation, focusing on commercialization of Class 8 trucks

As of this writing, AT technology developers are largely targeting SAE’s Level 4 autonomous trucks while the heavy-duty classes have increasingly garnered attention as the primary target for AT commercialization, particularly Class 8 dry van trailers. In terms of AT operations, although a safety driver is optional, the majority of these developers currently operate with a human driver onboard.

It is also anticipated that Level 4 autonomous trucks deployment will proceed incrementally, starting on highways in selected lanes where weather, regulations, and road infrastructure meet certain conditions. These initial deployments would likely be in the southwest states, particularly, Texas, Arizona, and New Mexico where weather issues are rare and the regulatory atmosphere is favorable. In fact, the majority of the testing and demonstration activities to date has taken place in these states. 

Emerging autonomous trucking scenarios

Given the state of AT technology, extant opinions accord that the initial commercial deployment of Level 4 autonomous trucks on public roads will center on long-haul trucking. In turn, a number of potential scenarios for AT implementations have been posited. A snapshot of these emerging AT scenarios is depicted in Figure 3. Essentially, potential long-haul AT implementations can be broadly categorized into two operating models, namely transfer-hub model and depot-to-depot model, as highlighted below.

Transfer-hub model scenarios. Also referred to as exit-to-exit model and highway-focused model, this operating model leverages a transfer hub through which an autonomous truck is deployed either alone or in conjunction with a regular truck to complete an origin-to-destination haul.

Depot-to-depot model scenarios. This operating model enables autonomous trucking without the need to leverage a transfer hub in which a self-driving truck is deployed to complete an origin-to-destination haul, either with or without direct human intervention.

 

Among these implementation scenarios, the transfer-hub model is currently adopted by the majority of AT technology providers (such as Embark, Kodiak, Torc Robotics, Waabi, and Waymo Via) to commence the first wave of AT deployment for long-haul transportation. Human oversight and remote assistance through teleoperations are also shared practices among AT players. Such approaches allow AT providers and users to deploy Level 4 autonomous trucks early, before transitioning to fully autonomous depot-to-depot applications when AT technology becomes more mature and its surface-street driving capabilities improve. Nevertheless, some experts opine that while teleoperations will be required during the transition period, remote human oversight may be a permanent element of autonomous trucking, even when it becomes fully autonomous.

Supply chain and logistics implications of autonomous trucking: A logistics-triad perspective

Potential SC&L implications of the wide-scale AT deployment depend to a large extent on application scenarios and levels of vehicle automation employed. Here, we explored potential impacts on SC&L management in reference to the following application scenario of focus: 

Level 4 autonomous trucks for long-haul freight transportation—including largely individual truck operations with possible hybrid truck-platooning implementations—under the transfer-hub model with human supports through teleoperations.

A logistics-triad perspective—involving shippers, transport service providers, and receivers—was adopted in this examination. The logistics-triad perspective underscores the intertwined relationships involved in both the exchange of goods and the exchange of transport services, stressing the embeddedness of transport activities in a broader supply chain network. Potential SC&L implications are summarized in Table 1 and further discussed subsequently.

 

AT implications on the exchange of transport services

When autonomous trucks take to long-haul highways, a number of implications for transport service providers can be perceived from two key aspects of the exchange of transport services, namely operating costs and service performance of trucking services provided.

Implications on operating costs

Top-three operating cost components for typical trucking services include driver costs, fuel costs, and truck costs, respectively. They constitute approximately 80% of total operating costs, about half of which pertains to driver costs, according to experts. The remaining 20% of total operating costs constitutes maintenance and repair (M&R), insurance premiums, tires and tolls, and licenses and permits. AT services under the transfer-hub model will significantly alter the structure of these conventional cost components, while bringing to the rank new operating cost elements associated with transfer hub facilities and teleoperations.

1. Changes in operating cost structure

Autonomous trucking’s largest cost impacts entail the two largest components, driver costs and fuel costs. In terms of driver costs, under transfer-hub AT services, the need for highway drivers is eliminated, limiting driver costs only to those related to off-highway operations. Concurrently, fuel cost savings can be achieved due to more fuel-efficient driving of an automated driving system, which can be programmed to follow best driving practices. 

The degree of these two largest cost savings will depend largely on the relative portion over the entire trucking route that is machine-operated (highway driving between transfer hubs) versus human-operated (off-highway driving inbound to and outbound from each transfer hub). That is, the larger the share of highway automation, the larger the potential savings. Also, by implementing a hybrid truck-platoon, potential fuel savings can be further enhanced for following trucks in the platoon as the front truck will reduce aerodynamic drag friction, rendering the followers the benefits of less air resistance.

Overall, decreased driver costs result in cost structure changes where the other major cost components—costs of fuel and autonomous trucks—will constitute larger shares of the total AT operating costs. The upfront capital costs of autonomous trucks can be significant since modern tractors, additional sensors, and a host of other CAV systems required could add $25,000 or more to the cost of a truck, according to industry experts. These costs, however, are generally decreasing with time as the technologies become more mature.

2. Mixed pictures on M&R costs, insurance and accident-related costs

As for cost components outside the top three, M&R costs as well as insurance and accident-related costs are widely recognized as areas potentially affected by AT implementations. Better safety performance of an automated driving system helps to assure safer operations and reduce risks of truck accidents caused by human errors. Consequently, savings in insurance and accident management can be achieved for AT service providers.

In terms of M&R costs, AT implications are not as straightforward, especially when both direct cost of M&R activities and indirect cost of availability loss are considered. On the one hand, advanced AT technologies can create M&R cost savings because mechanical wear and tear from driving and chances of accident-related damages are lessened by more optimized driving practices and better safety performance of autonomous trucks. Moreover, sensing technology and intelligent algorithms enable real-time information and more advanced fault diagnosis, providing more accurate information for maintenance planning.

On the other hand, new challenges and associated implications on M&R costs also arise due to increased truck utilization, complexity of AT technologies, and absence of a driver on the road. Since unmanned autonomous trucks are not required to follow federal HOS rules, they can be used more intensively, close to 24/7 continuous operating time. Hence, they would likely require more frequent maintenance, while simultaneously reducing time windows for planned maintenance. Due to this paradox, more unplanned maintenance could result, increasing uncertainties of direct maintenance costs and limiting the utilization of the truck (indirect costs of availability loss). Meanwhile, the absence of onboard drivers can further increase the need for predictive maintenance to minimize in-trip failures that would be difficult to address without a human driver. Also, the new, complex AT technologies might increase M&R costs as more skilled M&R workers and new or upgraded equipment are needed. All of these new challenges could offset any savings gained and negatively contribute to total M&R costs.

3. Key new operating cost elements: Transfer hubs and teleoperations

New incremental costs will be incurred in the provision of AT transport services, notably transfer hub costs and teleoperation costs. A transfer hub, which must be located in proximity of highway junctions, is a key feature in the emerging long-haul AT operations. It performs the crucial task of transshipment between autonomous highway hauls and non-autonomous off-highway hauls, and, when truck platooning is implemented, it can also perform the task of platoon formation. Furthermore, a transfer hub provides traditional terminal services such as refueling, on-site inspections and maintenance, as well as non-traditional AT data services related to data transfer offload for processing and storage. These transfer hubs will require labor, process coordination, and a relatively large space for parked trailers that have been decoupled from a truck platoon and are awaiting pickup, or for those trailers that await service on the tractor unit. The utilization rate and performance of transfer hubs will significantly influence the cost competitiveness of AT service operations.

Another key new operating cost element, teleoperation costs can be quite stiff. Compared to a traditional fleet command center, a remote AT center requires IT infrastructure and high-performance communication networks that enable the collection, processing, and sharing of autonomy-related data. It also requires more skilled operators who must possess a good understanding of AT technologies and how to use operating systems to, for instance, monitor autonomous trucks, update path planning, and initiate recovery behaviors to ensure that driverless trucks on the road are operating properly. These teleoperation costs, however, are expected to decrease over time as AT technologies become more mature. Besides, industry experts suggest that a skilled operator will be able to oversee as many as 10 to 30 trucks at a time from an office space—a sharp contrast to the number and work environment of human drivers required on the road to operate a similar size of truck fleet. 

Implications on trucking service performance

Transport service providers generally aim to reduce the lead-time length (time performance), improve lead-time consistency and safe delivery (dependability performance), provide visibility of shipment status (communications performance), and offer service flexibility in such terms as routing and delivery times (convenience performance). AT operations can potentially improve transport service performance in all of these dimensions.

At present, long-haul trucking operations are constricted by hours-of-service (HOS) regulations and the physiological capabilities of human drivers, limiting driving time to at most 70 hours and roughly 3,000 miles per week. Consequently, an average truck spends only 40% of its time carrying freight in the provision of services. In contrast, autonomous trucks are not compelled by these limitations, potentially allowing them to double the total distance covered in one day from around 400 to 600 miles to 800 to 1,200 miles. The productivity gained from increased truck utilization not only reduces operating costs, but also increases service capacity that avails more hours per day and more days per year of services for customers, provided that the trucks are properly maintained.

In effect, autonomous trucks not only can operate for longer hours without a need for a rest stop, but also includes the days of week and times of day that would otherwise be undesirable for human drivers to spend on the road—allowing them to travel farther, faster, and more flexibly in a given amount of time than a human-driven truck. Results are shorter delivery lead times, more consistent delivery lead times, and more flexible availability of services. Moreover, reduced traffic accidents due to the better safety performance of automated driving systems would greatly reduce service disruptions, while also ensuring reliable and safe delivery of goods.

Additionally, because autonomous trucks are connected in real time with highly precise positioning data required to safely navigate on public roads, in-transit visibility of the location and status of a shipment would also improve. Service performance can, hence, be enhanced by providing transport service buyers with accurate shipment tracking and access to real-time data about their shipments. Such visibility not only helps them better plan and manage loading/unloading tasks and other interfaces with the arriving truck, but would also make it possible for them to make timely and informed decisions in response to problems that may arise.

AT implications on the exchange of goods

To fulfill customer orders, shippers must make a series of interrelated logistics decisions and network design to effectuate the firm’s distribution strategies. The goal is to achieve efficient order fulfillment while creating satisfactory service levels for customers. Since transport services are the key influencer of order delivery performance, it is palpable that such a goal will be affected by AT implementations. Here, important strategic interactions in the distribution network that must be considered are transportation, inventory, and number and location of facilities.

Implications on inventory requirements

The inventory levels that a firm maintains at a facility in its distribution network impact the availability of goods it can dispatch to support order fulfillment. Typically, a shipper holds more inventory in order to shorten order cycle time. Meanwhile, transport services used for order shipping can differ in terms of transit-time lengths, transit-time variability, and damage rates; therefore, affecting both the absolute length and variability of order cycle time. By using faster and more reliable transport services, usually at a higher rate charged by transport service providers, inventory requirements and associated costs can be reduced.

However, AT long-haul services could rebalance the tradeoff between transport costs and inventory costs since not only do trucking services costs reduce, but time and dependability of the services also improve. The shorter transit time of order delivery would allow more frequent and smaller shipments that help to reduce both in-transit inventory and cycle inventory. Equally, more reliable transit time and safer delivery make it possible to reduce safety-stock inventory required to buffer against stockouts caused by lead-time uncertainties and unforeseen problems such as damage shipments.

Implications on number and location of facilities

The number of facilities in a distribution network can range from one large facility in centralized operations to several geographically dispersed facilities in decentralized operations. Generally, it is more economical to operate fewer, larger facilities due to the economy of scale, but firms must contend with the major drawback related to longer distances to customers, and resulting longer lead times and higher transportation costs. In contrast, having a larger number of facilities located regionally or locally allows firms to be closer to customers, enabling them to reduce order delivery costs and provide faster delivery. However, the value of better services gained are at the expenses of the additional costs of operating more facilities and costs of carrying more redundant inventory across the distribution network.

So far, to maintain suitable service levels, many firms have employed more decentralized distribution operations. However, as discussed earlier, with the coming of AT services, the one-day transit distance could dramatically increase and could double the delivery radius of a single facility. Such improvements would allow a firm to provide a comparable level of service, but with fewer facilities required. Another implication, given the transfer-hub model of AT services, is that proximity to AT transfer hubs could become a part of decision factors for facility locations to capitalize on the advantages of AT services.

Closing remarks

The recent decade witnessed the rapidly evolving nature of AT technologies that were once regarded as a conceptual envision but could soon become a reality. Prospectively, the first wave of commercial AT implementation is characterized by a transfer-hub model that leverages autonomous highway-driving supported remotely through teleoperations. Under such implementations, AT transport providers could potentially achieve significant operating cost reductions and service performance improvement that, in turn, provide SC&L managers with viable mechanisms to better effectuate freight transport, order fulfillment, and customer service. To this end, notwithstanding various barriers to broad-scale deployment—ranging from high initial investment, technological and infrastructural immaturity, to regulation inconsistency—AT potential benefits and revolutionary impacts on SC&L management are too great for businesses to afford indifference.

About the authors

Steve Tracey is a professor of practice in the Supply Chain and Information Systems Department within the Smeal College of Business at Penn State University. He can be reached at [email protected].

Kusumal Ruamsook is an assistant research professor at the Center for Supply Chain Research within the Smeal College of Business at Penn State University. She can be reached at [email protected].

SC
MR