Google reports 140,000 autonomous miles

Background

Google opened its 2010 post with a number: according to World Health Organization figures cited in the post, more than 1.2 million lives are lost every year in road traffic accidents ^[1]. The post argued that the technology it had been building had the potential to cut that number, perhaps by as much as half ^[1].

A self-driving car, in the general sense, carries sensors that build a picture of the road around it and software that turns that picture into steering, braking and acceleration. The point of the exercise is not to remove the human entirely but to move the human from doing the driving to supervising a machine that does it.

What was announced

In October 2010 Google published a post titled "What we're driving at" saying that its automated cars had logged over 140,000 miles, which the post described as a first in robotics research ^[1].

The New York Times, reporting on the same effort, put the fleet at seven test cars: the vehicles had driven 1,000 miles without human intervention and more than 140,000 miles with only occasional human control ^[2]. Those two figures describe different things — one a stretch of fully autonomous driving, the other a much longer accumulation of miles in which a person sometimes took over.

Where the cars went

The routes were chosen to look like ordinary driving rather than a closed track. Manned by trained operators, the cars drove from Google's Mountain View campus to the company's Santa Monica office and on to Hollywood Boulevard ^[1]. They also drove down Lombard Street, crossed the Golden Gate Bridge, navigated the Pacific Coast Highway and made it all the way around Lake Tahoe ^[1].

Between them those stretches cover city traffic, a famously steep and winding street, a major bridge, a long coastal highway and a mountain lake loop. For readers trying to judge how close the technology was to everyday use, the itinerary was as much of the story as the mileage total.

The safety record

According to The New York Times, the only accident in that record was when one Google car was rear-ended while stopped at a traffic light ^[2]. That detail is worth keeping in view: the one reported collision was caused by another driver, not by the software, and it happened while the car was stationary.

Why it mattered

For a robotics project, 140,000 miles on public roads was unusual, and Google's post presented the figure as a first ^[1]. It also changed the terms of the public argument. Instead of asking whether a car could drive itself at all, readers could ask how a fleet of them would behave across many miles of real traffic, and what a safety record built over that distance looked like ^[2].

The framing around the World Health Organization figures put the work in a public-health context rather than a gadget context: the measure of success was not a clever demo but a reduction in deaths on the road ^[1]. That framing — autonomy as a safety technology, judged by crash statistics accumulated over large distances — became the way automated driving projects described themselves.