Map applications are not new, yet more work is still to be done to form a detailed digital map of the roads across the planet. Moreover, the mapping roads is slow and tedious with standard technology. Even when aerial images are taken, mapping technology firms still need to spend many hours manually tracing out roads. This means there is an estimated 20 million miles of roads across the globe yet to be digitized.
How can this task be accelerated and made easier? The answer comes from the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory (CSAIL). This is in the form of RoadTracer, which is an automated method, built on artificial intelligence, and which can build road maps that are 45 percent more accurate than existing approaches.
The RoadTracer platform creates maps using a step-by-step approach. The software begins at a known location on the road, and then deploys a neural network to examine the surrounding area to determine which point is most likely to be the next part on the road. The plaform then adds that point and repeats the process to gradually trace out the road one step at a time.
MIT professor Mohammad Alizadeh explains: “RoadTracer is well-suited to map areas of the world where maps are frequently out of date, which includes both places with lower population and areas where there’s frequent construction.”
He adds: “For example, existing maps for remote areas like rural Thailand are missing many roads. RoadTracer could help make them more accurate.”
The following video shows the technology in action:
RoadTracer uses an iterative search process guided by a convolutional neural network-based decision function in order to derive the road network graph directly from the output of the convolutional neural network. When this new approach is compared with a segmentation method, as tested on fifteen cities, the error rate was just 5 percent.
