Self-driving cars: ‟There’s no such thing as zero risk”

By December 22, 2015
Luca Rigazio

[Part 1] The first part of our interview with Luca Rigazio, a specialist in self-driving systems and their interaction with humans, looks at how automobiles are evolving towards full autonomy, and the issues this raises.

Luca Rigazio, Director of Engineering at Panasonic’s Silicon Valley Laboratory, specialises in artificial intelligence and machine learning. Recently he has been focusing his research on automated systems – driverless vehicles and highly automated vehicles, robots and drones. L’Atelier met up with him to find out more about what the future holds for these technologies and the role they are likely to play in our daily lives.

You work on autonomous systems. Driverless cars are undoubtedly seen as one of the most spectacular innovations in the transport field. Do you think they will eventually be widely adopted by the public? If so, within what sort of time-frame?

Autonomous vehicles will certainly play an important role in the future. This will however take some time, not so much due to technical issues but because of all the questions around legislation, insurance, and social acceptance.  In my opinion, we’re talking about thirty years or so.  In fact the technology is almost ready now. We already have cars on the road that are largely autonomous – the vehicles now being sold by Tesla, for instance.

Image d'une voiture autonome de la marque TeslaTesla is already selling cutting-edge electric vehicles.

Tesla is today the most advanced player in autonomous cars, then?

The only cars on the road today that are able to self-drive are Tesla’s vehicles. As a startup, Tesla is able to take greater risks and get their cars to do more than models made by other manufacturers. For example, you can leave the car to drive all by itself while you text a message. That’s pretty impressive, but also worrying, given that the technology being used is less sophisticated than the technology of prototypes vehicles which are not yet on the market and still undergoing tests, such as the Google car. The Tesla car has far fewer sensors, for instance, so Tesla is taking a risk, in the spirit of Silicon Valley, to kickstart this new market.

What benefits do you think self-driving cars will bring us?

Well, human beings are never going to become vastly better drivers. Human capability has its limits. Driving ability can improve, but only marginally, and over several generations. Technology on the other hand is able to improve exponentially. So, sooner or later, autonomous cars will become better drivers than human beings, which will lead, among other things, to a reduction in the number of road accidents.

Driverless cars

Driverless and driver-operated vehicles: an unlikely coexistence?

Many people say driverless vehicles will have to attain zero risk status before they’ll be fully accepted by the public and can whizz around on the roads…

Well, there’s no such thing as zero risk! The possibility of error will always be there. I personally think that as soon as we have proof that autonomous cars cause fewer accidents than human drivers, initiatives will be set in motion to promote their development. It will simply be a matter of policy.  

Do you think that autonomous and traditional vehicles will be able to co-exist on the roads, or will the former entirely replace the latter? Will people actually be banned from driving cars in the future?

If you want to stop people from taking any action on the roads then you would not only have to ban them from driving cars but also from using bicycles and moving around on foot! And while it may not be unthinkable to give up driving your own car if that will reduce the number of accidents, who could possibly argue that we should give up cycling or walking? Unless you’re going to totally re-think the design of our towns and cities, I think it’s completely unrealistic to imagine a future where only artificial intelligences are allowed to use the roads. The basic challenge is to make artificial intelligence capable of understanding and predicting the behaviour of the human beings around them – with the element of uncertainty which that involves. Whenever we’re faced with a disruptive innovation, the most difficult time is always the transition phase.

Legal mentions © L’Atelier BNP Paribas