Autonomous trucks have gone through a classical hype cycle. After some bullish announcements early on and inflated expectations regarding the timing, players have changed their view on what is feasible in the mid-term. Slower-than-expected technology progress and safety issues have further pushed back these timelines. Now, it feels the fog is clearing and the roadmap, at least for the coming years, feels more stable.
The next milestone: ‘driver out’
Autonomous truck players have come a long way in training their systems on increasingly complex driving scenarios. Just five years back, autonomous drivers were capable of performing only simple capabilities such as maintaining lane position and performing lane changes; systems have since been developed to handle unexpected cut-ins, merges, night driving, further levels of inclement weather, and even construction sites. Many autonomous truck players are close to implementing the full set of capabilities that an autonomous driver requires or are currently solving the last remaining challenges, such as pulling over to the road shoulder or inspections.
From a total addressable market perspective, about 40% of all on-road freight traffic could be handled by autonomous trucks
The focus for most players is now shifting from expanding the capability set and the operating design domain of the vehicle to validating and improving their systems. Players will use 2023 to prove the safety capabilities of their technology. Safety incidents, like those seen in 2022, may impact this overall delivery timeline, but if things run smoothly, driver out operations can be expected in 2024.
While the ability to handle edge cases is critical for driver out operation, having a safe action as fallback, where the vehicle comes to a controlled stop in a safe location, is equally critical. In the early phase of adoption, we will see autonomous trucks sitting in a safe mode on the road shoulder while a remote operator in an on-ground support team takes over. While this will be feasible if the numbers of autonomous trucks in operation are small, the support needed, and thus the operating cost, must come down eventually. Driver-out autonomous trucks will not be immediately cost competitive with conventional trucks, even after accounting for the cost of human support. This proves that the driver out milestone is important, but theoretical. It determines the starting point of the market adoption curve, but the speed of adoption will be determined by both technological advancements and consumer behaviour.
Commercial adoption still far away
Parallel to technology development, autonomous truck players are working on commercialising their autonomous drivers. To accelerate market adoption, several players have chosen a business model where they act as carriers hauling freight for customers with their own fleet of autonomous trucks. The number of loads handled is still very low today but will grow as more customers sign on and routes are expanded. In the mid-term, autonomous truck players will also make their technology available to traditional truck carriers, likely through a pay-per-use subscription model.
Once the technology is commercially available, a large market potential can be tapped. From a total addressable market perspective, about 40% of all on-road freight traffic could be handled by autonomous trucks. This is the share of freight traffic that, today, is handled by combination trucks and travels a distance of at least 200 miles. Shorter routes will, at least initially, not be economical while trucks still operate in a transfer hub model with costly first and last mile drays.
Adoption will not happen overnight
Several near- and mid-term constraints will determine the speed of market adoption. While driver out operation is possible from a regulatory perspective in most US states, including Texas where freight volumes are concentrated, some states, especially California, specifically prohibit autonomous trucking, limiting short-term adoption potential. Also, as OEMs and technology players do not all operate on the same timeline, different levels of OEM readiness will result in staggered market entry, limiting vehicle supply.
If things run smoothly, driver out operations can be expected in 2024
Towards the end of the decade, market adoption will be determined more by the operational readiness of fleets and their risk profile. In general, tuck carriers are slow to adopt new technologies. This has been the case in the past with both telematics and lower-level advanced driver assistance systems, and will likely be the case with autonomous technology. Unionised fleets will face additional resistance. In addition, fleets need to change their operating model and integrate drayage runs into their operations to support the transfer hub model of autonomous trucking. While players will focus on high-density routes at the beginning, such as the Texas Triangle, routes with lower freight density will eventually need to be unlocked. To operate on these routes with the same ROI that players are used to from high density routes, technology costs must continue to decline. Finally, carrier consolidation is needed to drive market adoption beyond the industry’s large fleets and players.
Reaching full market penetration requires ‘unlocking’ many capillary routes, and eventually technology that is capable of handling dock-to-dock operations. Accounting for a quarter of the freight traffic by the mid-2030s would still mean rapid adoption, especially when compared to other similarly disruptive automotive innovations such as electric powertrains. It took almost three decades between the release of the first purpose-built electric cars in the mid-1990s and today’s full-scale production.
Walter Rentzsch is a Director at Roland Berger; Wilfried Aulbur is Senior Partner at Roland Berger
