Researchers at the Massachusetts Institute of Technology MIT have created a new model for enriching digital maps called RoadTagger that uses satellite imagery to define road features in digital maps, which may help improve GPS navigation, and the model is explained in a paper presented at the Advancement Association conference With artificial intelligence.
It is worth noting that artificial intelligence is said to be that maps are usually created all the time by means of the global positioning system (GPS) and that is by major companies such as Google, the global company known and specialized in this field alone, which sends vehicles equipped with cameras roaming within the neighborhoods to capture video Images of area roads, but this process is very expensive, and keeping these maps up to date takes a long time.
According to the high costs, some parts of the world are ignored, and only basic GPS data appears there, and one of the solutions to these challenges is to find a machine education model linked to satellite images, which is easy to obtain and update regularly to the extent What, in order to automatically define road features.
But the problem is that satellite imagery of roads is often mysterious because of things like trees and buildings, which makes things more difficult for the machine learning component, and that's where the new MIT innovation comes in.
The new model uses a mixture of neural network structure to automatically predict the number of lanes and types of roads (residential or expressway) behind obstacles, and the model in tests within isolated methods has been able to calculate lane numbers with accuracy of up to 77 percent, and it can deduce types of roads (residential or Rapid) with an accuracy of 93 percent.
is noteworthy that researchers are planning to enable RoadTagger to predict other features in many areas that are appropriate for everyone around the world, such as parking spaces and bike lanes through Google Maps, and researchers hope that maps will be used one day to help humans to Quickly validate continuous road adjustments with maps.
Sam Madden, co-author and professor in the Department of Electrical Engineering and Computer Science EECS and researcher in the Computer Science and Artificial Intelligence Laboratory (CSAIL) said: Most of the updated digital maps are for places that big companies care about, and if you are in places that they don't care much about, you are in A disadvantage regarding the quality of the map.
There is no doubt that artificial intelligence has only one goal and is to complete the process of creating high-quality digital maps to suit the unusual development currently in all parts of the world, so that they are available in any country.