Many innovations have a crucial graduation date: when they move from speculative R&D to real-world deployment and utility.
One of the more significant transitions underway is the rise of “autonomy as a service” (AaaS), a model where self-driving capabilities are not just embedded in cars but delivered as scalable platforms that are often entirely devoid of human drivers.
This, in turn, is giving rise to “driver-out” vehicles: machines that operate with no safety driver on board, signaling a bold leap toward fully autonomous logistics and mobility systems. Look no further for the tech’s graduation date than May Mobility’s recent partnership with Uber to provide thousands of vehicles with autonomous vehicle (AV) technology for proof of where the landscape could be heading.
“We know from the Waymo service that people are willing to wait longer for an autonomous vehicle. They’re willing to pay more, they prefer it so strongly,” Edwin Olson, CEO and co-founder of May Mobility, told PYMNTS. “And for companies like Uber, they know they need to be able to compete with that.”
For ride-hailing platforms, the economics of autonomy are compelling. Drivers, or more specifically, driver-associated costs, remain the biggest expense. Removing that variable opens up significant margins and also helps to eliminate variability due to labor shortages, hours-of-service regulations and human error.
Bringing the Driver-Out Experience to Life
The journey to fully autonomous ride-hailing is also about designing artificial intelligence (AI) systems that can handle the unpredictability of real-world road occasions.
“In the dark ages … there was a lot of just the intelligence of the programmer directly embedded within the DNA of the vehicle,” Olson said. “Now, AI methods are really the go-to approach for building autonomous vehicles.”
But machine learning on its own hasn’t been enough. Companies, Olson argued, have underestimated the volume of data needed to make AI systems truly reliable.
“Most AV companies have run out of cash before they’ve collected enough data to fully train those systems,” he said.
May Mobility’s own answer is what it calls multi-policy decision making (MPDM) — a kind of cognitive leap that adds reasoning capabilities to the autonomous stack. Rather than endlessly training a model on edge cases, MPDM simulates possible futures in real time.
“Imagine a forest of parallel universes,” Olson said. “In one of the universes, the pedestrian jaywalks. In another one, the pedestrian waits their turn.”
The system models these futures 1,000 times per second, scores each one for safety, comfort and legality, and chooses the best path forward. “We can do this reinforcement learning on the car in about 0.2 seconds,” Olson said. “Driving is much more like a dance than anything else. It’s all highly coupled.”
For incumbents and startups alike, the real question is not if autonomy will scale, but how to position oneself in a world where mobility is no longer just a product, but a service. Still, technology alone isn’t enough to win the race. Olson pointed out that scaling autonomy requires solving four interdependent problems: building vehicles, sourcing demand, managing fleets and mastering autonomy itself.
On the vehicle front, May has a deep relationship with Toyota, which builds the vehicles May’s technology powers. “Toyota can build all the vehicles we can eat,” Olson says, only half-joking. For demand, May is tapping into Uber and Lyft’s platforms.
Instead of convincing millions of people to download a new app, May’s driverless cars will show up right alongside human-driven ones in apps people already use.
Autonomous Adoption Won’t be the Hard Part
One of the most surprising lessons from May’s journey, Olson said, is how willing people are to try autonomy. Especially if it makes their lives easier.
“There’s often a sentiment expressed that, oh, you know, adoption’s going to be a real barrier. And here, I actually just totally disagree,” he said.
His evidence? Human behavior. “If you went back in a time machine 15 years ago and said, ‘Would you get into a car driven by a complete stranger?’ people would say no. But then you land in San Francisco and you need a ride … and suddenly it’s fine,” Olson said.
Autonomous vehicles, he argued, follow the same pattern. “If you can solve a real-world problem, make their lives more convenient, they’re going to adopt this.” That doesn’t mean safety isn’t paramount — “It’s one of our absolute pillars,” he said — but public trust, he believes, comes through utility first.
And if the future sounds like science fiction, Olson would remind you: it’s already on the road today.
