If there is no person physically driving the car it is called Autonomous Driving. Then it is important to know who is actually driving, i.e. who is responsible of the act. It can also be that some type of semiautonomous driving is an option: a combination of physical driver and computer-aided system (either in the car or in the infrastructure). It is evident that travel time is radically shorter in the case of autonomous driving. It is important that there are good systems for e.g. speed control.
It is important to test the reactions of the car (how well the car can function). The driving wheel could have a heart beat monitor and other life indicators, thus ensuring to avoid bad drivers: drunken, old, sick, young (risky drivers!).
The speed control can be done by cameras using a black box in the car. The "outside" signals coming from infrastructure might do the speed regulation, behaviour control, witnessing the happenings on the road and in the car, driving in the background all the time and recommending breaks, mandatory overnights etc.
Increasing the perception of risk with acoustic design of vehicle and acoustic risk indication features could be developed. Systems can be developed to avoid also adverse and bad weather driving, and intelligent road condition monitoring. The liability questions must be solved, too.
The intervehicle communications (and catering for vehicles that do not have such communications) is central to this Idea. CVIS is a project that attacks one part of this, but the Idea at the second stage is to determine standards ond communciations systems that will allow autonomous and human driven vehicles to share roadspace.