The service, which will include a human behind the wheel in case the autonomous vehicles malfunction, marks a significant milestone for Waymo, a company that began as a secretive project within Google in 2009.
Since then, its cars have robotically logged more than 10 million miles on public roads in 25 cities in California, Arizona, Washington, Michigan and Georgia while getting into only a few accidents — mostly minor.
The company is initially operating the new service cautiously, underscoring the challenges still facing its autonomous vehicles as they navigate around vehicles with human drivers that do not always follow the same rules as robots.
The service, dubbed Waymo One, will only be available to a couple of hundred riders at first, all of whom had already been participating in a free pilot programme that began in April 2017.
It will be confined to a roughly 100-square-mile area in and around Phoenix, including the neighbouring cities of Chandler, Tempe, Mesa and Gilbert.
Although Waymo has been driving passengers without humans behind the wheel in its free pilot programme, it decided to be less daring with the new commercial service.
“Self-driving technology is new to many, so we’re proceeding carefully with the comfort and convenience of our riders in mind,” Waymo chief executive John Krafcik wrote a blog post heralding the new service.
The ride-hailing service is launching in the same area where a car using robotic technology from Uber hit and killed a pedestrian crossing a darkened street in Tempe, Arizona, seven months ago.
That collision attracted worldwide attention that cast a pall over the self-driving car industry as more people began to publicly question the safety of the vehicles.
The Uber robotic car had a human safety driver behind the wheel, but that was not enough to prevent the accident in March.