What Can We Learn From Autonomous Vehicles’ First Driving Test?

Over the past three years, commercial trials for autonomous taxis have commenced in several cities in the US and China. Companies such as Cruise and Waymo in the US and Apollo in China are allowing consumers to order and pay for pickups by autonomous ‘robotaxis’ via an app. Regulatory approval for these trials has come in phases, broadly following the pattern in Figure 1 below: 

The trials have scaled up over the past three years, with Apollo (part of tech giant Baidu) active in at least nine Chinese cities and recently announcing the addition of 200 more AVs to its fleet by the end of 2023. In the US, Cruise (backed by General Motors) and Waymo (a Google subsidiary) have expanded to cities across the Southwest and serve customers with a combined fleet size of roughly 1,000 vehicles. Between September and November 2022, Cruise reported more than 2,500 paid passenger trips, covering over 26,500 miles. 

In the US, firms have adopted different ‘depth’ vs ‘breadth’ strategies. Firms like Cruise and Waymo have opted for depth and are running more sophisticated passenger-charging trials across a limited number of locations. Other AV firms like Motional (a joint venture by Hyundai and Aptiv) have emphasised breadth and are running earlier stage trials across a wider range of locations and conditions. 

Trials reveal new jobs for humans

The trials have been broadly successful, and firms are continuing to expand into new locations. Apollo recently announced that they will offer services in Hefei, China, and Cruise is planning to offer paid trips in Austin, Texas, later this year. Public uptake has been strong; there are currently large waiting lists to gain access to Cruise’s robotaxis in San Francisco, Austin, and Phoenix, Arizona. This reflects a gradual long-term shift in public perception of the safety of AVs, which is steadily improving (see Figure 2).

Accidents resulting in injury have been very rare, but there have been frequent reports of minor disruptions, such as AVs causing congestion by halting in the middle of the road when encountering roadworks. These growing pains suggest that even when AV taxis become more widespread, they may not be able to operate entirely autonomously, with some human roles still required: 

Engineers — to be deployed to vehicles which are experiencing defects; this could range from a technical issue with a sensor to a broken window or seatbelt 

Fleet operators — to remotely take over navigation of a vehicle in emergency situations; this is already being done by Apollo in China 

Customer service reps — to handle customer complaints, which could range from payment issues to lost property requests 

Cleaners — to maintain the vehicles’ appearance 

Several hurdles remain before fleets of robotaxis become commonplace in major cities 

Technology 

The robotaxi trials are a step towards widespread AV adoption, but significant barriers remain. AV technology has advanced over the past few years, as evidenced by the improving disengagement (i.e. disengaging from autonomous mode) data reported during testing on California roads. However, more testing in a greater range of inner-city environments is required. 

Thus far, testing in European cities with older, narrower, more irregular road layouts has been limited. This could prove more challenging than the predominantly grid style city layouts where most AVs trials are currently taking place. AV technology also requires more testing and trials in more diverse weather conditions. The advanced American trials have taken place in cities with predominantly warm, dry climates — Cruise’s license in San Francisco does not currently permit it to operate in heavy rain. AV technology still needs to demonstrate that it can perform in extremely wet, snowy and stormy conditions before it can realistically receive widespread regulatory approval. AV firms recognise this, and Waymo has announced plans to trial their AVs in Bellevue, Washington, to collect performance data in a predominantly rainy climate.

Regulation

A number of local governments have allowed AV testing on public roads; however, most countries still lack detailed legal procedures to account for fleets of level 5 AVs sharing the road with human drivers, and all the associated complications.¹ Some AV companies like Waymo and Aurora are seeking to engage regulators to discuss how such legal systems can be developed. Waymo recently published a document which they described as a blueprint for developing a wide-ranging safety case to be delivered to regulators and have encouraged other AV firms to contribute.

Stimulating consumer demand

The final barrier to widespread AV adoption is stimulating consumer demand. Improving overall public confidence in AV safety is part of the challenge, but another factor is that many consumers must also be willing to forgo the enjoyment of driving. This trend is already underway (see Figure 4), and there could be an initial opportunity for robotaxis to capture inner-city journeys, where traffic is significantly more congested and driving is less enjoyable for motorists. 

The other aspect to stimulating consumer demand will be lowering the cost of robotaxi trips so that they are cheaper than ‘traditional’ taxis and a viable alternative to public transport in some cases. Robotaxi firms will have to pay for the production or leasing of AV fleets, in addition to potential labour costs associated with operating the vehicles (such as engineers, customer support reps, cleaners, etc). To become price competitive, they will need to ensure that these costs are offset by the large potential savings of not having to pay human drivers. This is still many years away, as the mass production of advanced AV technology remains expensive. 

Once these barriers are overcome, what’s next? 

The commercial trials are a notable milestone for AV technology and have provided valuable early learnings on public perception and how robotaxi fleets may operate in the future. They have also shown us that there are still significant hurdles to clear before we move into the fast lane towards widespread AV adoption. 

However, if these hurdles can be overcome, we can start to examine how AVs may enter the market on a wider scale. 

For more information, please contact strategy@lek.com. 

L.E.K. Consulting is a registered trademark of L.E.K. Consulting. All other products and brands mentioned in this document are properties of their respective owners. © 2023 L.E.K. Consulting

Endnotes
¹Per SAE International’s definition of automation levels, 5 is “total vehicle control with no intervention under all conditions.”