X

MapLite software developed by MIT could speed autonomy in rural areas

The system is able to navigate without the need for hyper-detailed 3D maps, making it attractive for use outside of urban centers.

Kyle Hyatt Former news and features editor
Kyle Hyatt (he/him/his) hails originally from the Pacific Northwest, but has long called Los Angeles home. He's had a lifelong obsession with cars and motorcycles (both old and new).
Kyle Hyatt
2 min read

The dream of an autonomous future is creeping closer every day, but one large problem is the fact that today's self-driving car technology relies heavily on hyper-detailed 3D maps in order navigate on roads. Currently, nearly all of these maps are of major cities, with mapmakers and autonomous car developers being seemingly reluctant to work on mapping more rural areas.

Researchers at MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have developed a type of autonomous driving software that navigates based on sensor input and basic GPS data, like what you'd find in a navigation app on your phone. The system is called MapLite, and it's reportedly shown great promise in testing.

argo-ai-princeton-lidar
Enlarge Image
argo-ai-princeton-lidar

MapLite uses lidar sensors in combination with an IMU and basic map data to provide self-driving functionality on rural roads.

Argo AI

In addition to a lack of resources being dedicated to mapping more suburban and rural areas, often lighting, signage and road markings are less than ideal, which proves challenging for most current self-driving cars to parse. One advantage though is that the roads are often less complex and less populated, meaning that there are fewer things for the software to deal with.

"The reason this kind of 'mapless' approach hasn't really been done before is because it is generally much harder to reach the same accuracy and reliability as with detailed maps," says CSAIL graduate student Teddy Ort. "A system like this that can navigate just with onboard sensors shows the potential of self-driving cars being able to actually handle roads beyond the small number that tech companies have mapped."

MapLite uses lidar and inertia sensors to help predict road conditions up to 100 feet in front of the vehicle and checks that against the basic GPS data to help guide it. The system is significantly less complex than others and could be a way to speed up autonomous car adoption outside of urban centers if it's pursued.

MapLight is currently being tested in Devens, Massachusetts, in a modified Prius, and the results have been positive. MapLite is still in development and has limitations related to issues like elevation changes, but it will be exciting to see how it develops in the near future.