The conversation about autonomous vehicles has been hard to ignore over the past few years. Indeed, it’s been almost deafening at times, and yet there is still a perception that the technology is a solution in pursuit of a problem. The reality is that, for many parts of the world, mass adoption of autonomous vehicles is unlikely over the medium term.
In North America, however, things are different, and autonomous trucks are on the way. Fleet trials are already ongoing, and our expectation is that these trials will result in true commercial application by the middle of this decade.
On a longer-term basis, we expect adoption rates for autonomous trucks—again, in North America—to run out at somewhere between 20% and 40% by 2035. This is a development that will be hugely disruptive across the entire trucking value chain, and it is one that all stakeholders—fleets, truck makers and component suppliers—need to be both aware of, and to have strategies in place in order to be ready to cope.
The argument for autonomous trucks is already compelling, and looks likely to become more so in a relatively short timeframe. Legacy industry participants ignore it at their peril
Why North America? The answer is, for a combination of reasons. Long-haul distances in the US are much greater than in Europe, at approximately 850km (528 miles), versus approximately 130km, making it a more attractive use case. Secondly, outside of urban areas, highways are much less congested and weather conditions, at least in areas like the Sun Belt, provide fewer challenges to autonomous driving systems.
Moreover, favourable legislation is already in place in many states in the US to allow the commercial testing and operation of autonomous trucks on public roads without requiring the use of an attendant driver. In some regions—especially along the I-10 corridor—several states already allow the commercial operation of driverless trucks, and a clear path towards broader adoption is now visible.
While there are some reasonable concerns still regarding societal acceptance of autonomous trucks, they too are likely to reduce over time. Many states do not differentiate between testing and deployment today, and only require a safety self-assessment. However, this is about to change, with states demanding independent certification audits for autonomous systems. Technology has also progressed considerably in recent years, and Level 4 trucks are already operating in confined areas such as mines and ports today; by 2025, we expect this technology to be ready for interstate operation. Finally, hardware costs are falling—for sensors, LiDAR and ECUs in particular—and will see further reduction through scale effects once the technology is being adopted.
We are seeing two approaches in the market. The first is to define a narrow ODD and make the system work for this first, and then expand the scope; the second involves aiming for a broader ODD from the outset. Both approaches have their justification
While Level 2 (partial automation) systems are already becoming standard features for high-end trucks, we expect fast progression from here to Level 4 (high automation) technology. This allows driverless operation in certain situations, such as highway driving. The fact that viable business models can be built around Level 4—meaning that full automation will not be required—makes autonomous driving a more imminent development in the trucking industry than in the passenger car industry.
We see many players in the market working on operating models for Level 4 truck, from technology start-ups to truck manufacturers. The key difference between the models is their Operational Design Domain (ODD). The ODD describes the specific conditions under which an automated truck is intended to function. It defines where a system is designed to operate (e.g. what roadway types, geographical areas and speed range) and when it is designed to operate (e.g. environmental conditions such as weather, daytime/nighttime, season). The broader the ODD, the more complex the system becomes, meaning more testing and validation of the algorithms is required. This drives up cost and development time.
We are seeing two approaches in the market. The first is to define a narrow ODD and make the system work for this first, and then expand the scope; the second involves aiming for a broader ODD from the outset. Both approaches have their justification: while the first can be seen as a minimum viable product that can be brought to market quickly, and by doing so helps to develop the market, the broader approach aims at a solution that is ready for commercial implementation, but at the cost of higher development expenses and a longer timeline.
We expect larger fleets to be among the early adopters, followed by mid-sized fleets over time. Smaller fleets—especially those in the owner-driver segment—are likely to be excluded from this trend
As the operating cost savings of autonomous trucks are likely to be significant—we estimate up to a 40% reduction—we anticipate a pull effect from the fleets. While fleet testing is already happening, we expect broader commercial adoption starting in the middle of the decade. This will most likely go through stages, with the I-10 corridor acting as nucleus. From there, other high traffic trucking routes will be developed.
In terms of fleets, we expect larger operations to be among the early adopters, followed by mid-sized fleets over time. Smaller fleets—especially those in the owner-driver segment—are likely to be excluded from this trend, as they will not have the financial means to invest in the technology; given that they will also, as a consequence, not be able to benefit from the operating cost savings, this segment is likely to come under greater consolidation pressure. With the addressable market being mainly the large and mid-sized long-haul fleets, we estimate that over the long-term, 20-40% of the Class 8 truck population could become autonomous.
Autonomous trucks will have multiple industry implications. The higher productivity of driverless trucks that can—theoretically—operate 24/7 will result in much higher asset utilisation. The type of trucks will also change, with fewer sleeper cabs sold than today. As Level 4 trucks will become cost competitive to rail-intermodal, freight volumes will be shifted from rail to road.
One of the core competencies of a fleet today is to provide a safe driver. With autonomous trucks, this capability becomes obsolete
Ignore this at your peril
In the long-term, the whole industry structure is at risk of being disrupted. One of the core competencies of a fleet today is to provide a safe driver. With autonomous trucks, this capability becomes obsolete. Supply and demand matching, another key capability of a fleet, can be automated by digital freight matching systems. And lastly, the importance of own maintenance capacities will deteriorate with predictive maintenance. New players that provide capacity-as-a-service (CaaS) directly to the shippers might emerge and bypass the large fleets. There are several potential contestants that could become the CaaS service provider, ranging from large fleets themselves, to rental companies, technology start-ups or even truck manufacturers. Ultimately, a scenario is possible in which the industry splits into autonomous long-haul trucking, served by CaaS providers, and driver-based regional haul, dominated by the traditional fleets.
This is not the first time that the trucking industry has been faced by disruption, but this time it is different. The advent of new technology here has levelled the playing field and traditional legacy industry participants are now confronted—arguably for the first time—by competition from start-up operations many times smaller than themselves.
We believe that it would be foolish to ignore these start-ups on the basis of either size or relative newness to the trucking industry. The argument for autonomous trucks is already compelling, and looks likely to become more so in a relatively short timeframe. Legacy industry participants ignore it at their peril.
About the authors: Dr. Wilfried G. Aulbur (email@example.com) is Senior Partner and Member Supervisory Board at Roland Berger; Oliver Dixon is a Senior Advisor to Roland Berger