Driverless cars or autonomous vehicles may be just around the corner, but in order to get there, they require a meticulously mapped environment that goes beyond the turn-by-turn directions on a GPS.
According to Nidhi Kalra, information scientist at the RAND Corporation, automakers are pretty much universally interested in developing driverless cars, but they do not all have access to the kinds of map technology they need.
Driverless cars require maps made for machines, not humans. Kalra explains that “for an autonomous vehicle a map really is not a map; it’s a blueprint of the world. Where’s the curb? Where are the lanes? Is that a stop sign or a stop light? Where are the constructions? Where are the detours? It’s not a map anymore the way we use it; it’s something different, it’s much richer.”
Acquiring maps of this quality means huge investments for automakers in the infrastructure, technology, and servers that can deliver precise information at high speeds. It is unfeasible for many to develop their own mapping technology because the maps difficult to create and costly to maintain.
Kalra emphasizes the stakes of designing quality maps: “The technology needs to be able to distinguish between something that’s inconsequential and something that matters. The decisions that need to be made are extremely hard. And every vehicle that has this technology or uses it is participating. And it’s absolutely essential. The difference between being 10 centimeters one way or another way can be the difference between being in the gutter or being in the other lane into oncoming traffic. So, there’s not room for error with this technology.”
While there are many autonomous cars that can drive by themselves, responding to the world around them, making split-second decisions, and staying out of the gutter is another task entirely.