According to Martani: “Railway stations and airports in world cities from London to Beijing come under enormous pressure during peak times, when overcrowding in confined spaces can put passengers' safety at risk. Being able to predict congestion ahead of time means we could avoid overcrowding, making public transport safer and more efficient.”
Counting millions of people as they move rapidly around cities and predicting where they will move next, is a major challenge – one that Martani and his colleagues at CSIC are studying in the corridors of the Department of Engineering at Cambridge.
He installed two depth sensors in the corridor which, linked to Counterest® software, counts staff and students as they cross a virtual line in either direction. And the latest test results are promising, showing the system is 85% accurate, which Martani is now working to improve.
“The most difficult conditions for the counting software seems to be associated with people moving their heads and hands rapidly, so by improving both the software and the calibration we should improve the system reliability for predicting crowd movements ahead of time,” he says.
He's now taking the system out of the Cambridge corridor and testing it in more realistic conditions at London Bridge station.