To Drive Is Human

We perform an enormous number of functions while driving, and a computer trying to match a human driver must face all of them. Cornell University researchers offer an insider's perspective on the issues the DARPA Grand Challenge competitors faced in creating a humanlike driver - without the human. In preparing for the Grand Challenge, Cornell University's team divided the driving problem into three basic tasks: (i) localization - knowing where you are, (ii) sensing - seeing what's around you, and (iii) path planning
_e
termining how to get to a destination. Creating a vehicle capable of driving itself therefore required finding computational solutions to each of these three tasks