This article has been reviewed according to Science X's editorial process and policies. Editors have highlighted the following attributes while ensuring the content's credibility: College students are already under stress from classes, tests and presentations. So when something as "easy" as finding a parking spot is anything but, it can cause a bad start to an already challenging day.
With that in mind, a team of George Mason University students developed an app that uses data and artificial intelligence (AI) to remove the guesswork and save students from those interminable trips to the top of a parking garage, only to find it's completely full. Parkeye predicts parking availability, helps users find lots that match their permits and provides information about parking changes and closures. The app was created by computer science students Tona Otoro and Sami Saleh, with support from computer science students Shabeer Manalai, Trong-Tin Mai and Rafsan Islam, and Aayusha Sapkota from the Costello College of Business.
The idea grew out of a familiar student experience: arriving on campus without knowing where to park. "Everyone's kind of having this similar problem," Manalai said. The team surveyed students around campus and found that many struggled with everything from finding open spaces to understanding parking regulations.
The team then approached George Mason's Parking and Transportation office, which provided five years of parking data that became a foundation for the app. "We frequently partner with students on class projects, and what made the Parkeye team unique was that they pursued this project independently, outside of a formal class or research program," said Josh Cantor, director of Parking and Transportation. "We applaud the students' initiative and innovative approach; their crowdsourcing model reflects the kind of creative thinking that can contribute to future parking technology solutions.
We are evaluating a variety of approaches, including crowdsourcing, camera-based technologies and space sensors to help improve the parking experience for our campus community." The app considers about two dozen factors when predicting parking availability, including historical data, class schedules, weather, the distance between lots and buildings, and the number of classes taking place during particular time periods. As more people use the app, the team can incorporate additional data and retrain its models. Saleh said the initial model achieved approximately 91% accuracy against the parking data provided by the university.
The developers also have testers visit parking areas to compare actual availability with the app's predictions, using those results to continue refining the model. And the app does more than predict whether a lot is crowded. Users can filter parking by space type, including electric vehicle charging spaces, and receive updates about lot closures and other changes that could affect parking.
Extract — continue reading at the source.
