Uber has just launched its driverless (automated) taxi service in Pittsburgh, Pennsylvania, in the US. Though the company isn’t the first to roll out such taxis—that distinction belongs to Singapore’s
nuTonomy—its fleet of self-driving cars will contend a far more complex and larger traffic and geography than the Singaporean service, given Pittsburgh’s undulating terrain and maze of roads, alleys and flyovers; nuTonomy has taken on a much smaller area in a flat city that is famous for its well-planned roads network. Uber’s hesitation in letting it be a completely hands-off affair, at least till kinks have not been identified and sorted out, is evident in the fact that the company will put two technicians—one to monitor the car’s overall performance and the other to take on the driver’s role at sticky spots—for now. It plans to bring it down to one technician, still behind the wheel, before switching to fully automated driving.
Against the backdrop of the widely-covered accident involving a Tesla self-driving car, in which the lone occupant was killed, such caution seems quite justified. But Uber, unlike a Tesla or a Google or an Apple (all of which are either developing or have rolled out self-driving cars), is still better-geared to undertake the phased self-driving experiment. In Pittsburgh, as also the rest of the world where it has operations, the company has a rich mine of data on road and driving conditions and grid complexity—thanks to the billions of miles the regular Uber taxis have already clocked—that it can analyse to fine-tune self-driving technology and real-time responses in the real world.