Question: How long do you think it will be when all/most cars on the road will be fully autonomous?

Giuseppe Cotugno answered on 6 Nov 2017:

That’s a very good question, I won’t be able to give you a precise timeline though. I won’t expect a Vauxhall Corsa to be able to drive itself very soon, but I would see an expensive Mercedes being able to drive for a limited period of time if the whether is good. For completely autonomous cars we will have to wait a little while.

This question deserves a longer explanation. In my fair opinion, the technology to allow a car be driven by itself under specific conditions (i.e. good visibility, good lightening etc.) are nearly there. What need to be improved are the sensors: either they are too imprecise or too expensive or the sensor used are too specific (i.e. mostly normal cameras and ultrasound sensors are used, but a lot more things can be used: infrared cameras, touch sensors etc.). Another thing that requires improvement is the clever and quick processing of a lot of data from multiple sources (many different sensors) so that the car can have a better idea of what is happening around it. This can help the car driving on more difficult whether conditions for longer, since the longer.

What I think is the biggest limitation to overcame at the moment is the human/legal factor. When you drive, accidents happen because of a distraction, or uncontrolled emotions or a wrong estimation of the behaviour of any of the drivers involved in a crash. Computers cannot get distracted or angry but their behaviour might not be logical and this can anger people (if you ever died in a first person shooter because a friendly team member thrown a grenade at the wall bouncing back to you, you known what I mean). If a car’s behaviour in case of danger is not made predictable and human drivers do not know how autonomous car normally drive if they have a problem (the car’s version of the “running against the wall” behaviour of videogame’s AI) emotional drivers will likely to run into accidents and oppose the innovation. This open the door to the even bigger problem of deciding who’s the responsibility of a crash with an autonomous car, especially if the car did not crash because of an error in the algorithm but because a human driver behaved wrongly in a moment when the software suffered from lack of information (say the sun flashed the camera for a fatal second). Engineers will have to be very clever to avoid such dangers and lawyers and policy makers should be wise enough to help engineers doing their job the best and to inform the population on how such cars work and which are the risks.

I am sorry for this extra long answer, I hope it was informative other than verbose.