either have the driver in complete control, or have complete autonomous function (weather full driving or just parking) mixing will just lead to tragedy.

Right. If I had a nickel for every time a stability controller or lane-keeping system led to tragedy... wait...
I really like MIT's idea. Humans have proven they're good at decision making, yet bad under pressure. Computers are not as good at decision making, yet good under pressure. Combining the two seems like you could exploit some serious synergies.

Good idea for when it is human error causing the incident, when it is a fault with the car that caused it, I would query how much of a good idea it is to rely on the car to know how to react.

According to this US Department of Transportation study, 99% of the traffic crashes investigated were caused or contributed to by human error. (http://www.nhtsa.gov/people/injury/research/UDAshortrpt/). For as much as people like to blame the vehicle (eg. Toyota accelerator crisis), human error is almost always the source of the problem.

Observing an intersection in any major American city (and many large cities around the world) would prove that humans simply don't belong behind the wheel. The only reason traffic isn't completely anarchic is that people still have an instinct for self-preservation. Otherwise, most drivers break as many laws as they think they can get away with.

re; DaveG People get used to not driving when they are driving and then get into a different car.

How the system will identify whether it's better to 1. hit the object (which suddenly started to run over the road) or 2. hit another car but avoid object? How the system will know whether this object is child or dog/chimp/pig etc.?

Have we considered the cost of the Velodyne LIDAR sensor? At about $75000 a sensor, how many of you are willing to add the sensor to your cars? While I respect the research efforts, I think that academia must keep the biggest factor in mind when it comes to product improvements: Affordability.

I've been saying for almost a decade that autonomous control is the way to go for cars if cars are to remain our primary vehicle for day to day travel. But I like driving. This idea is great if you ask me.

re; warren52nz
Fully autonomous cars will be great but encouraging drivers to give less of their attention to their driving while they are driving is a bad idea.