A network of sensors that is used at stations to combat unsafe situations during busy periods also appears to be suitable for checking whether travelers are complying with the 1.5-meter distance.
A team from Eindhoven University of Technology came to this conclusion in a study in the scientific journal Plos One.
For three years now, cameras have been hanging above the head of an unsuspecting public from the ceiling of three platforms at Utrecht CS, Schiphol and Amsterdam-Zuid. They shoot images, after which the computer calculates the distance to the head, shoulders and then the rest of the body. A unique “depth fingerprint” is created in a fraction of a second. After which the computer can track the movements of each individual and record their walking route.
Rail operator Prorail is already using the system to control crowds and to set up stations more safely and efficiently. But the ultimate goal is “to gain a fundamental understanding of how people behave in large crowds,” says researcher Alessandro Corbetta from TU / e’s Faculty of Applied Physics.
Due to the corona crisis, lockdowns and working from home, there have been no crowds or large crowds in public transport for months. But the network also appears to be suitable for seeing whether the people who do travel keep sufficient distance from each other.
Recognize households
To do this, the system had to learn a new trick. After all, it should not work if couples or family members from the same household are too close to each other. For this, the researchers wrote an algorithm that recognizes possible family-like groups very precisely and at the same time. The idea is that two or more people who enter the station together, stay together during their journey through the station and eventually get on the same train together presumably belong together.
The data does not contain information that can be traced back to a person, the researcher says. “The camera image is discarded as soon as the computer has made its height calculations.” The network of nineteen cameras that has been hanging above the platform of track 5 in Utrecht since 2017, now takes 100 thousand measurements every 24 hours.
“With better information, we can communicate more specifically to our travelers, or pull off some benches or stairs where too many people pass. But we are also working on high-tech solutions, such as automatic speakers that can focus on busy places to ask specific travelers to walk a bit, “says Frank van Schadewijk, data analyst at ProRail.
Improve furnishing
In the longer term, simulations can help improve the layout of existing stations. “Before the corona outbreak, we assumed a 30 percent growth in train traffic over the next ten years. Stations are often located in the middle of busy city centers, so with hardly any room to expand, “says Van Schadewijk. “Then you need to know how to manage passenger flows in this limited environment.”
The system could also be used in museums. “Because of the available space, it is sometimes difficult to keep to a distance of five feet and you also want to be able to distinguish couples or families,” says Corbetta. According to him, the system can be scaled up to other places where tens of thousands of people pass, such as at an airport.
