The Driverless Paradox: Why the Robotaxi Needs a Better Municipal Curb
- Apr 26
- 3 min read
Author: Bern Grush Date Published: April 26, 2026
There's a familiar argument in industrial AI circles: the smarter the software gets, the more the bottleneck shifts to physical infrastructure. The brain outruns the body. A parallel paradox is now unfolding on city streets—except the body in question is the curb.
Autonomous driving systems are getting measurably better. Fewer collisions. Fewer injuries. Waymo's self-published safety data and independent assessments continue to show performance advantages over human drivers across current operating domains. Level 4 ADS is unlikely ever to be perfect—no technology is—but it is becoming demonstrably more reliable and safer than the humans it replaces within its operational design domain.
That progress deserves acknowledgment. It also reveals something uncomfortable.
As the machine improves, the remaining deficits in the full robotaxi experience become harder to attribute to the machine. They increasingly trace back to something the machine cannot fix: the curb it arrives at.
The curb doesn't know the vehicle is coming
A robotaxi that navigates flawlessly across miles of urban traffic but has no safe place to stop when it arrives, now blocks a firetruck or straddles a bike lane. A passenger who uses a wheelchair may need to enter or exit a robotaxi while it stands in the through lane. A blind traveler might be dropped off on the other side of the street forcing an unnecessary and risky street crossing. An ambulance carrying a critical patient cannot get through a queue of autonomous vehicles waiting for passengers exiting a theatre.
None of these failures are ADS failures. They are curb failures—failures of unmanaged, uncoordinated, dumb public infrastructure meeting a vehicle class that requires something smarter.
Shifting blame, shifting responsibility
Here is the paradox: the better—and cheaper—the robotaxi becomes, the more visible these curb failures become. And the more they will be blamed, unfairly but understandably, on the robotaxi industry.
A passenger whose wheelchair cannot reach the vehicle, because the accessible zone is occupied by a standing sedan, does not file a complaint with the city. They file a complaint with the operator. A commuter who is late for a meeting because their robotaxi circled the block searching for a place to stop will not admire the vehicle's flawless driving. They will question the technology's fitness for purpose.
Public acceptance of robotaxis will not be determined by crash statistics alone. It will be determined by the full experience—from the moment someone opens an app to the moment they step out at their destination. The curb sits in the middle of that experience, and right now, as the actual experience of riding in a robotaxi becomes increasingly uneventful, is occupied by scrolling. Now the critical, eventful trip experience is getting in and getting out. And this is rapidly becoming the most important measure of safety, comfort, and convenience.
Mismanaged, the end points for pickup and drop-off (PUDO) are already generating congestion. It will only get worse.
Accessibility is not optional
Safe vehicle access is a baseline expectation—but it cannot be guaranteed without dedicated space at the curb. Those spaces must meet ADA-level accessibility standards. This is not aspirational, it’s an obligation. But any robotaxi system that cannot guarantee a safe, unobstructed, accessible PUDO spot—for wheelchair users, for visually impaired riders, for elderly users, for people with cognitive disabilities who rely on spatial predictability—cannot honestly claim to improve mobility for those who need it most.
This is critical. The populations most underserved by existing transportation—people who cannot drive, who cannot use fixed transit, who live in transit deserts—are precisely those who stand to gain most from autonomous mobility. That longstanding promise of the driverless vehicle is degraded every time the curb fails them.
What needs to happen
Cities that host or plan to host robotaxi operations need to treat curb management as a critical infrastructure function, not a parking enforcement afterthought. That means reserved, dynamically managed PUDO zones. Real-time vehicle-to-curb coordination. Accessible space guarantees. Priority protocols for emergency vehicles. Enforcement mechanisms that actually clear blocked zones.
This is not science fiction. The standards frameworks exist. The technology to orchestrate it exists. What is missing is municipal will—and the recognition that the robotaxi's social license depends not just on how well it drives, but on what it arrives at.
The robotaxi industry is delivering a capable brain. Only cities can provide a better body.
For more on addressing this issue: https://curbreserver.com/download.html?doc=case-study
Other publications by the Author:
The End of Driving (Edition 2) Automated Cars, Sharing vs Owning, and the Future of Mobility
By Bern Grush, John Niles and Andrew Miller https://shop.elsevier.com/books/the-end-of-driving/grush/978-0-443-22392-1
Comments