Although you may never have seen it happen yourself, it isn't all that uncommon for large objects - including people - to fall onto the tracks at subway or railway platforms. While security personnel viewing CCTV feeds will catch some of these accidents, the cameras' shots are sometimes obscured by people, poor lighting, or even the trains themselves. The results can range from lengthy delays in rail service, to fatalities. Now, however, researchers working on a project for the Université Lille Nord de France have developed a system that uses radar to automatically detect and identify objects that fall onto the tracks. When installed at a platform, the system could then shut off power to the tracks, and notify oncoming trains.
The system continuously sends out wideband radio waves, and analyzes their reflections when they're bounced back by a foreign object, via an Automatic Target Recognition procedure. Only the most prominent features of the object's reflected signal are processed, and then compared to a database of known objects.
In a computer simulation, the researchers tested the system using objects such as suitcases, bottles, and the bodies of human adults, adolescents and children. In all cases, it was able to accurately identify the objects. Physical tests were also conducted in an echo-free chamber, in which the radio waves were guided towards two men, a woman, and two pieces of luggage made of different materials. Once again, it was able to differentiate between the subjects.
"We hope these devices will be used in the near future since they are very complementary to existing video systems and have a similar final cost," said Ali Mroué, lead author of a paper on the research. "The complementary use of video and radar systems could lead to low levels of false detection, which is mandatory for this application, and maximize the chance of survival for passengers who have fallen on the line."
The paper was published today in the journal Measurement Science and Technology.
See the stories that matter in your inbox every morning