Self-Driving Cars

Ars Technica has a very interesting series on self-driving cars. Part I discussed the technical challenges to implementing the concept. Part II discusses the societal effects of self-driving car adoption. Part III will talk about the public policy, political, and legal issues this all would raise.

I've thought about this for a few years now. Particularly when stuck in traffic or on an overnight road trip. There is nothing about driving, particularly highway driving, that is too difficult to train a robot to do. Highway driving seems simpler than speech recognition, which you can do on your home computer, for example.

City driving is certainly more complex, but as the DARPA Urban Challenge showed, it is within reach. Plus it seems that in these difficult driving environments, additional information could be used. Think of Google Street View. If Google can take pictures of all the streets in most major American cities just for giggles and a few ad dollars, surely someone could develop a visual concordance for use in automated vehicles.

Long-haul trucking, in particular, seems the ideal entry point for self-driving vehicles. If a trucker could run their trucks 24 hours a day and pay no labor costs, that would be a significant advantage. Plus a convoy of a dozen automated trucks could drive inches from one another saving fuel and taking up less space on the freeway.

Part II of the Ars Technica article brings up the concept of driverless taxis. Imagine your car drives you to work in the morning, then takes off to sell it's services as a taxi during the day, refueling as needed and making you money while you work. Then in the afternoon it returns home with you. Or imagine tiny cars that delivered pizza from the local pizzeria, a gallon of milk or a dozen eggs from the grocery store.

This would be transformational, but it is also feasible.