字幕表 動画を再生する 英語字幕をプリント [MUSIC PLAYING] Maneuvering a vehicle in any type of weather can come with its own set of challenges and limitations. Maneuvering a vehicle through conditions that limit visibility, such as mist or fog, can be even more challenging or even dangerous. But now, thanks to a team of researchers out of MIT and their newly developed system, there may be a solution to this problem. MIT researchers have developed a novel imaging system that can gauge the distance of objects shrouded by fog so thick that human vision can't penetrate it. An inability to handle misty driving conditions has been one of the main obstacles to the development of reliable autonomous vehicular navigation systems. So the MIT system could be a crucial step toward self-driving cars. To test their system, the team placed objects in an enclosed box approximately one meter long and then gradually filled the space with thick fog. Outside, pointing into the box, there is a laser which fires pulses of light into the foggy scene and then a camera that measures the time it takes their reflections to return. What they found was their system was able to image objects even when they were indiscernible to the naked eye. More specifically, in fog so dense that human vision could only penetrate 36 centimeters, their system was able to resolve images of objects and gauge their depth at a range of 57 centimeters. 57 centimeters is not a great distance, but the fog produced for the study is far denser than any that a human driver would have to contend with in the real world. The vital point is that the system performed far better than human vision, whereas previous systems have performed worse. The system is designed to get around the issue of light reflecting off water droplets in fog, which confuses most imaging systems, making it almost impossible to discern objects ahead. The MIT researchers developed an algorithm that uses statistics about the way fog typically scatters light to separate the raw data from the camera into two parts, the light reflected from the shrouded object and the light reflected from the fog. The light reflected from the object is then used to image the scene and calculate the object's distance. Of course, visibility is not a well-defined concept, since objects with different colors and/or textures are visible through fog at different distances. So to assess the system's performance, the team used a more rigorous metric called "optical depth," which describes the amount of light that penetrates the fog. Optical depth is independent of distance, so the performance of the system on fog that has a particular optical depth at a range of one meter should be the same as its performance on fog that has the same optical depth at a range of, say, 50 meters. In fact, the system may even fare better at longer distances, as the difference between light particles' arrival times will be greater, which could make for more accurate images.