Deep Learning Technology Predicts Road Accidents
According to studies, by combining historic accident data, road maps, satellite images, and GPS with a machine-learning model, we could be closer than ever to safer roads. Over the years, technology has evolved a great deal. GPS systems have eliminated the need for memorizing streets, cameras and sensors that alert us to objects near our vehicle, and even autonomous electric vehicles. The precautions that we take while driving have largely remained unchanged. Most of the time, we rely on road signs, trust and hope to reach our destination.
Researchers at the MIT Computer Science and Artificial Intelligence Laboratory, in collaboration with the Qatari Center for Artificial Intelligence, have developed a deep-learning model to predict high resolution maps of accident risk. The model predicts the number accidents for a future time period using data from past accidents, road maps and GPS traces. The map can identify high-risk areas and future accidents.
According to reports by homelandsecuritynewswire.com, maps of this type have been captured so far at much lower resolutions, resulting in a loss of vital information. The previous attempts relied mainly on historical crash data. However, the research team has collected a large base of information and identified high-risk zones by analyzing GPS signal data that includes data on traffic density. speed, direction, as well as satellite imagery which provides data on road structure. The researchers found that highways were more dangerous than residential roads nearby, and that intersections and exits from highways were even more dangerous.