Possible, is another issue.

Three weeks ago I posted my skepticism about the idea that driverless cars could ever expand their territory, like human drivers can. Today, I have found a good summary of how complex the mapping part of such an endeavour be in “Surprising Benefits of Maps for Autonomous Vehicles (AVs)", which I invite you to read in full.

That post explains that “while safety is the first goal of maps for autonomous vehicles, there is so much more they can enable”, including:

• understand well in advance when, for example, to take curves, in order to minimize consumption (of fuel, tires, brakes…) and disturbance for passengers
• reduce car sickness, by issuing subtle sounds before and during a turn, to let the passengers know that it is happening (see details in the post)
• pick the right lane to avoid potholes
• location of safe places to pull off the road if the AV has problems
• location of all crosswalks, to let vehicles know “where to pay proper attention to the pedestrians, even on the sidewalk"
• location of all the places where it is _“OK to U-turn or 3-point turn”, instead of carrying people on much longer, useless paths

This looks great. Will it be actually doable, affordable, and sustainable in the long run, since all the stuff above needs continuous checks and update of the maps? No idea.

Image source: art from a DeepMap announcement