^ I don't know how much driving you do in snow, but Canadian winter driving is very.... variable, and unpredictable. I wouldn't expect it to be easy to model in software, by any stretch. I'm not saying it's impossible, but I can see lots of ways it would go wrong. On top of that, even in summer, our roads are very often in terrible shape - the potholes and cracks everywhere are probably easier for software to avoid, but it's still a big obstacle. IMO, the biggest thing automated driving is going to lack is a sort of intuition. Things like understanding when it's appropriate to take an alternate route because school has just started back, and there's construction starting on a particular route, and certain roads are going to be cracked for a few months after winter because of the ice before they can fix it, or how to change up the drive because of heavier foot traffic in some areas at certain times of the year, or how to get around an accident or something that there's no data for, etc. I've always thought that automated transportation would only really work on a large scale once we're able to build dedicated infrastructure for it. I can imagine a whole bunch of cool futuristic automated transportation systems that would maybe work really well if we have infinite resources to implement them. But the current roads and cars weren't made with automation in mind.