A popular physicist joke is nuclear fusion is thirty years away and always will be.
While robo-taxis are picking up passengers today in San Francisco, Austin and Phoenix, robo-taxis seem perpetually a few years away.
We’ve had military aircraft drones deployed since 1995, a modified autonomous Volkswagen vehicle won the 132-mile DARPA Grand Challenge in 2005, and six automakers announced in 2015 delivery plans for their self-driving vehicles between 2017 and 2020. One of those companies announced yesterday:
General Motors (GM) will slash spending in its self-driving car unit Cruise, after an accident last month seriously injured a pedestrian and prompted regulators to retract its operating permit for driverless cars in San Francisco.
In October, the company said it would no longer operate its vehicles without safety drivers behind the wheel.
The horrific accident in San Francisco highlighted a significant challenge for autonomous vehicles, “long tail” or edge cases. These instances are at end of the distribution curve of occurrence and are often unique. Robo-taxis can operate perfectly for ten thousand miles then break down with an edge case. AI requires many training examples to learn. Humans are more flexible and leverage common sense to navigate these cases. Another challenge for autonomous vehicles is social acceptance. While humans learned to live with over forty thousand deaths from car accidents per year, it is too early to know what will be accepted from autonomous vehicles.
For more, see GM Slashes Spending on Robotaxi Unit Cruise, a Setback for Driverless Cars
Recent Comments