Robotaxi

Self-driving cars had been a technological aspiration since the DARPA Grand Challenges of 2004-2007, which proved autonomous vehicles could navigate desert courses and urban environments. Google's self-driving car project, launched in 2009, demonstrated that the technology could work on public roads. But the gap between demonstration and commercial deployment would take over a decade to cross.

Waymo (spun out of Google in 2016) launched the first commercial robotaxi service in Phoenix, Arizona in October 2020. Passengers could hail fully autonomous vehicles—no safety driver—through an app, just like Uber or Lyft. The service initially operated in a geofenced area of suburban Chandler, gradually expanding across the Phoenix metro region. By 2024, Waymo had expanded to San Francisco, Los Angeles, and Austin, completing millions of paid rides.

The adjacent possible for robotaxis required multiple technological streams to converge. Lidar sensors provided 3D environmental perception. Deep learning enabled object recognition sophisticated enough to distinguish pedestrians, cyclists, and vehicles in complex urban scenes. High-definition maps created precise localization. Massive compute power—both in-vehicle and cloud-based—processed sensor inputs in real-time. Perhaps most critically, millions of miles of testing data allowed systems to encounter and learn from rare edge cases.

The regulatory environment proved as challenging as the technology. Arizona's permissive approach attracted Waymo's initial deployment. California, despite hosting most autonomous vehicle development, imposed stricter requirements. China pursued a different path, with companies like Baidu, WeRide, and Pony.ai deploying robotaxis in designated zones of Wuhan, Beijing, and Shenzhen, often with government support that would be unusual in the US.

The path to robotaxis was marked by setbacks. Uber's fatal pedestrian accident in Tempe, Arizona (2018) led to industry-wide reassessment of safety standards. GM's Cruise, after expanding aggressively in San Francisco, suspended operations in late 2023 following regulatory pressure over incident reporting. The technology worked, but edge cases—construction zones, emergency vehicles, unusual behavior—continued to cause problems.

Geographic concentration shaped development. Waymo emerged from Google's Mountain View campus. Cruise operated from San Francisco, backed by GM's Detroit manufacturing. Tesla's approach—using cameras rather than lidar, training on data from millions of customer vehicles—came from Palo Alto. Chinese robotaxi development centered in Beijing, Shanghai, and Shenzhen, reflecting the country's tech hub geography.

By 2025, robotaxis had proven the concept but not yet achieved the scale that would transform urban transportation. Waymo expanded steadily but remained limited to specific cities. The economics were improving—autonomous vehicles could operate 24/7 without driver wages—but the capital costs of sensor-laden vehicles and extensive mapping remained high. The future seemed inevitable, but the timeline remained uncertain.