Should driverless cars kill their own passengers to save a pedestrian?

[IATIA]

... I agree that this is not some moral issue.  Like the train track dilema.  But this one fails to give me much of a dilema.  The car will do what it is programmed to do, which will be to try and avoid killing anyone.

As you say, "The car will do what it is programmed to do", but that is where the moral issue resides. There is a programmer or team of programmers, that must consider the outcome of the decision making process that will ultimately control the vehicle. I do not know if there are any real programmers on board, but programming must encompass worst case scenarios. In the case of computer software and games, generally the rule of thumb is that any input which fails to align with the intended programming is just shoved in the bit bucket and the software will resume polling inputs. However, in the case of cars and planes, it would be disastrous to ignore unforeseen inputs, so worst case scenarios must be considered, albeit they can usually be grouped into similar algorithms, but that is where the problem comes in. Simple algorithm, something in way, dodge it. OH NO! Cliff, too late. It does not matter how fast the computer is because it must abide by physics and causality, i.e. reason for stopping and stopping distance. Even though it can calculate, for sake of argument, the exact stopping distance, it is still obligated by the laws of physics. This does not consider tire wear, that patch of oil (hit one on a bike once, no fun) or any other unknowns. Because the car cannot think and can only do what is programmed, ultimately the programmer has to consider the decision to program the safety of the occupants or the safety of the greater numbers or whatever. This is where the morality lies and ultimately the obligatory lawsuits. It is obvious from some of the posts, that some posters did not really read the article linked at the beginning. So yes, from the programmer's stand point, this is exactly the trolley problem.