Follow
Subscribe

Autonomous Cars Can Get By With a Flash Drive-Sized Map, Says MIT

Home > News > Content

【Summary】Highly detailed maps take up a lot of storage space. MIT is developing a new system called MapLite, which uses a minimalist map that could fit on a flash drive. Self-driving cars can reference this map, and then use their sensors to navigate to a destination.

Mia Bevacqua    May 12, 2018 8:00 AM PT
Autonomous Cars Can Get By With a Flash Drive-Sized Map, Says MIT

Before GPS, humans would reference a map, then use their senses to find an exact location. Researchers at MIT say self-driving cars could do the same thing. 

Mapping the world on a flash drive

Most autonomous vehicles rely on highly detailed digital maps – often provided by Google – for navigation. The problem is, these maps need to be astronomical in size to cover everything. Plus, roadways are continually changing, requiring data to be updated often.

"Maps for even a small city tend to be gigabytes; to scale to the whole country, you'd need incredibly high-speed connections and massive servers," says Teddy Ort, a graduate student in robotics at the Massachusetts Institute of Technology's Computer Science and Artificial Intelligence Laboratory. "But for our approach, a global map could fit on a flash drive."

MIT's reasoning is that self-driving cars don't need maps with precise detail. Instead, these vehicles can use a crude map and their sensors to navigate the rest of the way. The institute is testing this theory with its system called MapLite. 

MapLite in action

So far, MIT has only tried MapLite on Massachusetts backroads. The test subject was a Toyota Prius modified with lidar and other sensors. Not surprising since the Toyota Research Institute backs the project. 

Many other researchers use machine learning for navigation. MIT says it only uses this tactic to find what road the vehicle is on. 

"We do use machine learning to find what road it is," Ort says. "But our path finding is all from a model-based approach. If it doesn't work as we thought, we can go in and fix it." 

So far, the main drawback to MIT's approach is verification. There's no crew of researchers out driving and documenting the route – only a very smart vehicle. 

"The main conceptual drawback is verification," Ort says. "A detailed map means someone's driven over it, done a fair amount of testing and shown that it's safe—it hasn't changed. But if you've never driven over it before, that's not so. We're working on how to verify the safety of driving on a road we've never seen before."

Yeah, that could be tough. But if directionless humans can do it, there's a good chance autonomous cars will be able to do it also. 


Source: IEEE Spectrum

Prev                  Next
Writer's other posts
Comments:
    Related Content