Maths, science and computer algorithms are combining to help drivers find the perfect car parking spot close to their destination.
A Charles Sturt University research team, headed by Sabih Rehman, has designed and developed a mobile phone app that uses advanced machine learning and historical parking data to predict the availability of car parking spaces at any one time. Machine learning is the study of algorithms in computer systems that rely on patterns to gather data.
“It’s really exciting to see how accurate the predictions are and when you come back and compare them with the real-world observation,” Dr Rehman said.
“Say if you are travelling to attend an important business meeting, you can find a parking space close to that venue with almost 75 to 80 percent accuracy.”
Easing traffic congestion
Circling to find a car parking space accounts for 30 per cent of traffic congestion in major cities. The NRMA said predictive technology like this could only help to reduce that statistic.
Spokesperson Rebecca Page said a recent survey found more than four out of five Sydney motorists admitted to avoiding certain locations altogether because they were scared of not being able to find a car park.
“Almost half went back home because they couldn’t find a park; so this is telling us that parking is a really big problem for people,” she said.
“If we can get people into parking spaces quicker that means they’re not driving around for as long, which means we’re also reducing congestion.”
Machine learning is the future
Smart technology like parking prediction is expected to grow and become more prevalent.
“The Internet of Things [IoT] revolution is starting to take hold globally, with billions of sensors and devices all connected via the internet; it has the power to solve so many problems,” Dr Rehman said.
“The same principles applied to the parking problem can be applied almost anywhere, from improving a manufacturing process, or monitoring people’s health, to making our cities more liveable.”
Dr Rehman said the technology was still a work in progress and the team would now trial new algorithms to get accuracy up to 90 per cent.
