Waymo's Rough Patch: Freeway Restrictions, Construction Confusion, and a Miami Collision
This week brought Waymo three separate reminders that the last mile of autonomous driving is also, often, the messiest. On Monday, the company confirmed that one of its robotaxis was struck by a vehicle that drifted through an intersection in Miami over the weekend, releasing dashcam footage of the crash. Days earlier, California's Department of Motor Vehicles restricted Waymo's freeway operations in the state, citing specific safety incidents and warning that the company's highway driving posed a "hazard to public safety." And in a separate report, Business Insider detailed how Waymo's vehicles continue to stumble through construction zones, a persistent weak spot that has dogged the company even as it has expanded into new cities.
None of these episodes, on its own, is catastrophic. The Miami collision, according to Waymo's own account, was caused by another driver, not its software. The construction-zone confusion is a known limitation, not a new one. And the DMV's freeway restriction is a regulatory tightening, not a shutdown. But stacked together in a single week, they read less like bad luck than like a pattern: a company whose technology performs impressively in the controlled chaos of city streets but still flinches at the unscripted chaos of the real world — cones placed without warning, lanes that vanish mid-block, a car losing traction at 45 miles an hour.
That distinction matters enormously for an industry trying to convince regulators, insurers, and ordinary commuters that driverless cars are ready for the highway, literally. Freeways are where autonomous vehicle companies most want to prove themselves, because higher speeds and longer trips are where robotaxis could eventually outcompete human-driven ride-hailing on cost. A state DMV publicly invoking the phrase "hazard to public safety" is the kind of sentence that gets quoted in every subsequent hearing, lawsuit, and city council debate about whether to let these cars expand.
Construction zones are a more mundane problem, but mundane is exactly the issue. Robotaxis don't need to solve rare, dramatic edge cases to earn public trust; they need to handle the boring, daily friction of American infrastructure — the orange barrels, the flagger waving traffic through with a hand signal no camera has been trained to parse. Waymo has spent years arguing that its disengagement rates and safety data outperform human drivers on average. Construction zones are where "on average" runs headlong into specifics.
For riders, the stakes are personal rather than statistical: does the car you're trusting with your commute know what to do when the lane it was mapped to drive in no longer exists? For the industry, the stakes are existential in a slower-burning way. Waymo remains the furthest along of any robotaxi operator, which also means its stumbles are the ones regulators, competitors, and the public scrutinize most closely. A rough week for Waymo is, by extension, a rough week for the premise that driverless cars are simply a matter of scaling up what already works.