Assisting Drivers Using Parking Predictions Through an Automobile App

Traffic and parking are becoming increasingly unbearable in Malta, with delays on Maltese roads being almost triple the European average [1]. For this research, we have taken the parking problem at the University of Malta (UM) as a case study. UM has approximately 720 parking spaces available for a total of 11,670 students. The main parking area is usually full by 7:20 am but does experience changes in parking when students finish class, which were the focus of this work. This dissertation is a part of the SmartSkyAutomator project and presents research on three main areas of study: data science, pervasive computing and mobile applications. The aim of this study was to create a solution by combining these areas of research, that makes the process of finding a parking space more efficient.

Following an observation study, the model for the study was created. The upper part of Car Park 6 at UM was used in this study and the days taken into consideration were Mondays, Wednesdays and Fridays. For each of these days, the 10:00 am and 12:00 pm time intervals were studied i.e. a few minutes before to a few minutes past the hour. The use of a commercial drone helped to build a dataset based on the model, and different Regression algorithms were tested on this dataset. The best one overall was selected to make parking predictions. The vehicle detection tool

in [2] was used simultaneously, in attempt to obtain identical values to the manual logs of the dataset, obtaining satisfactory results. After designing three prototypes, an automobile app using web technologies and a Node.js framework was built to give predictions stored in a MongoDB database to drivers. A controlled experiment was designed to evaluate the solution. This involved eighteen drivers using the automobile app in a real-life scenario to find a vacant parking space. A usability questionnaire was then answered to evaluate the usability and safety of the application.

The outcomes of the experiments showed that finding a parking spot was the hardest on Mondays, whilst Fridays were the easiest. Additionally, it was easier to park at 12:00 pm than 10:00 am. The questionnaire revealed that participants found the app simple, effective and safe to use. Drivers preferred using the buttons on the touch screen rather than voice commands to interact with the app. The app achieved a very high overall score. With more collectors, several parking areas can be studied at one go and the study can be extended to include other car parks. The experiment would be more realistic if a larger dataset is collected to include data for other car parks, rather than creating a random dataset for them. The best algorithm for predictions would have to be reselected based on the new data.


[1]         T. Bonnici, “Study shows that delays on Malta’s roads are almost triple the European average”, The Malta Independent, 30-Jan- 2015.

[2]         S. Mallia, C. Attard, and R. Farrugia, “Automatic Vehicle Detection from Aerial Imagery”, University of Malta, 2018.

Student: Andrea Naudi
Supervisor: Dr Conrad Attard
Co-Supervisor: Dr Ing. Reuben Farrugia
Course: B.Sc. IT (Hons.) Software Development