With the recent headlines vaunting the success of self-driving vehicle projects from California to Stuttgart to Mannheim to Parma, people are starting to wonder: how long until I can buy a self-driving car? It turns out that the answer is not so simple.
This fundamental question came up during the opening lecture of the V-Charge summer school on self-driving cars, which took place in Zurich the week of July 7, 2014. Dr. Steven Shladover pointed out that, before automated driving enjoys wide acceptance, we will have to demonstrate that it makes driving safer. How difficult would it be to demonstrate this?
In the EU, there is approximately one vehicle fatality for approximately every 120 million kilometres driven, as Stevev Shladover sustained in his presentation during the summer school. Find more info on accidents statistics. 120 million! If we use 10 or 100 test vehicles, imagine how long would it take to show that our self-driving cars are better than that? The challenge seems hopeless until we realize that the numbers above include all driving—in cities, on the highway, on rural roads…everywhere—and we can make room for fully automated vehicles if we simply reduce the scope of the problem.
This is precisely the approach of the V-Charge project: it focuses on enabling valet parking and charging of electric vehicles. Valet parking is a great application for automation: the cars drive at low speeds, highly detailed maps of the parking area can ease the burden on the vehicle perception system, a central server can manage parking and charging stations and, best of all, drivers have the convenience of never having to worry about finding a parking spot. It sounds great, right? I think so, too. And that is why I’ve spent the last three years working with an outstanding team of university and industrial partners to address the scientific challenges posed by this application. We’ve even got a smartphone app that will call the car to pick you up.
So: how long until you can buy a self-parking car? In my opinion, it may be sooner than you think.
Please note that these are only my opinions, they don't represent the opinions of the V-Charge project.
This argument is based on Steven Shladover’s talk at the summer school (slides available here), which updates the calculations derived in . The accident statistics for the European Union are from . Because  doesn’t include the number of crashes that resulted in fatalities, I was unable to reproduce the calculations in  exactly. However, the argument still holds.
Shladover, S.E., “Cooperative (rather than autonomous) vehicle-highway automation systems,” Intelligent Transportation Systems Magazine, IEEE , vol.1, no.1, pp.10,19, Spring 2009
European Transport Safety Council, “Ranking EU Progress on Road Safety, 8th Road Safety Performance Index Report”, June 2014 http://etsc.eu/wp-content/uploads/ETSC–8th-PIN-Report_Final.pdf