Roadshow

MIT tells self-driving cars: Take me home, country road

Self-driving cars currently struggle with country roads, but not for long.

SF Motors self-driving test vehicles
SF Motors

Self-driving technology is still in its infancy and there are still questions to answer

Among those questions: What happens when you take a self-driving car out of the city, off the freeway and start driving on country roads?

Country roads are problematic for a number of reasons: They're poorly marked and pretty difficult to map, particularly when you consider there's little incentive to map them in the first place because so few people drive on them. So what are the options? We could switch back to regular driving like neanderthals, or we could get to work on a whole new set of technologies to make country driving easier for self-driving.

That's what MIT is currently working on

It's called Maplite, a framework that allows self-driving cars to adapt on the fly, to drive on roads they've never driven on without 3D maps. It combines Google Maps GPS data with LIDAR and IMU sensors, that use pulsed laser lights to identify distances. This allows for a safer self-driving experience on country roads.

There are still kinks. For example, Maplite struggles with mountain roads, mainly because of the elevation. Considering the stakes are much higher when you're driving at incredible heights, that's an issue, but it's a problem MIT is hoping to solve. 

You can find out more about the research here.