The Real Prospects of L4 Driverless Tech

The news about Level 4 autonomy has been mixed. In May, Waymo announced that it was doubling its robo-taxi service area in Phoenix to 180 square miles, with the goal of increasing its current 10,000 trips a week tenfold.

On the other hand, the decision by the California Public Utilities Commission to allow the 24/7 commercialization of L4 Waymo and Cruise robo-taxis in San Francisco led to numerous mishaps, including a Cruise vehicle crashing into a fire engine and another Cruise robo-taxi getting stuck in wet cement. These and other incidents moved the Commission to ask Cruise to reduce the number of its self-driving vehicles in the city by half. Waymo experienced few such incidents, no doubt because of its successful experience of collecting fares with its robo-taxi fleet in the Phoenix area since December 2018.

Though there are no numbers, these robo-taxi fleets on the streets of two large cities suggest that at the moment there may be more L4 vehicles in operation than L3.  Notably, Waymo gave up on the L3 business model in 2017, after some drivers fell asleep during testing and were unable to take control of the car when they had to and General Motors, which owns Cruise, doesn’t have any L3 vehicles on sale to the public yet.

Bryn Balcombe, autonomy systems and regulation expert at Oxa (formerly Oxbotica), said the availability of both L3 and L4 illustrates the current situation of the AV industry, which he described as: “What’s the cost of the system that achieves the commercial benefit?”

Mercedes-Benz, which is selling L3 vehicles in Germany, is always looking for the price point that the consumer will pay, he explained. “These [robo-taxi] companies that are launching are doing so on the back of investments; they’re not doing it on revenues they are generating.” L3 vehicles are affordable because they operate in a simple operational design domain (ODD). “You don’t have people on the road, you don’t have cyclists on the road, you have all the traffic going in one direction and you have ample space to the side. A Waymo car could drive in that environment 100% but would a consumer pay for all of the sensing and computing technology that Waymo has developed on a car that they would buy in a showroom?”

Oxa is currently providing its software to several Level 4 projects, such as the global tech company ZF, for use in passenger shuttles that will operate 24/7 in fixed routes in cities around the world. ZF currently has six autonomous shuttles operating in a segregated lane, with fencing on both sides to keep out pedestrians, in Rivium, a business park in the Dutch province of South Holland. The service was launched in December 2022. Its next project is to take this service to a mixed traffic environment, like a bus traveling in a lane on a trafficked road, including the usual risks of that domain, such as pedestrians, cyclists and traffic lights.

“They are working with Oxa to get to that next level,” Balcombe said. This will require more complex sensing and computing technology. “The segregated lane mode is very similar to the L3 mode, it’s a limited domain,” he explained. In a mixed-traffic ODD “there is more complexity, and you need to have more redundancy.”

The focus for Oxa, he said, is “in what industries will it happen first? Where is the commercial need that actually covers the cost of the system at the moment?” Much of the company’s early interest was in off-highway use cases, such as mining, airports and the solar farms, wind farms and petrochemical plants Oxa is engaged in with BP. “These are industrial applications where this technology has a commercial benefit immediately,” Balcombe said. “The cost benefit of doing something in mining is very different from doing the same thing for robo-taxis. The robo-taxi is probably the hardest commercial business case.”

He believes that L4 technology will be first widely used in business applications. “Consumers at the moment will buy goods off Amazon; that has been solved,” Balcombe said. The problem is to reduce costs on getting those goods to the consumer as quickly as possible. “Robo-taxis, which is a consumer application will come later, and we think that you buying a vehicle with L4 technology will come a lot later.”