It might take a few years until car makers are willing to take on the liability of decisions their cars make and override a driver's preference but a model developed by MIT engineers gives a good idea of how such a system may operate.
MIT's "intelligent co-pilot" distinguishes between safe zones and all other driving areas. If a driver leaves a safe zone and a system detects "hazards" in the vehicle's surroundings and the co-pilot takes over until the car is back in a safe zone. Hazards would include stationary objects that the vehicle may collide with on its current course. In an article, MIT said that the system allows a driver to control the vehicle, only taking the wheel when the driver is about to exit a safe zone." The innovative thought is "shared" control between a driver and a vehicle's computerized controls that relies on the detection of surroundings by cameras, infrared sensors, or laser-based systems.
It is an evolutionary approach as today's systems are far from being able to take entire control of a vehicle. Much of the problem lies with the limited field of view of a vehicle that cannot slow or accelerate with the consideration of other cars that are too far to the left or to the right of the car. Accordingly, the MIT researchers said that their "team has run more than 1,200 trials of the system, with few collisions." Collisions apparently occur when "glitches in the vehicle’s camera failed to identify an obstacle. "In a conclusion, MIT said that "for the most part, the system has successfully helped drivers avoid collisions."
"For the most part", of course, is not good enough since we do not know if the system would actually cause collisions the driver could have avoided. Auto-pilots will only be able to take over control of a car when they will provide 100 percent protection from accidents. Volvo may be among the companies that are very well aware how these systems can fail.