Design and Development of Object Detection System in Augmented Reality Based Indoor Navigation Application (Case Study: Faculty of Mathematics and Sciences Building, Syiah Kuala University)

The Faculty of Mathematics and Natural Sciences building in Syiah Kuala University is a three-story building which consists of many rooms. The number of rooms causes the process of finding a room will take a lot of time. Currently, the informatics team in Syiah Kuala University is developing an application which implements Indoor Positioning System. One of the features is an indoor navigation feature based on Augmented Reality (AR). However, the route from one point to another is not always straightforward. Therefore, this research will design and develop an application which aims to detect obstacles present in the route from starting point to ending point. Obstacles that can be detected in this application are 3D objects, classified as human, bag, chair, potted plant and table or desk. In the application, TensorFlow Lite is used to detect obstacles in the form of 3D objects, where the distance is calculated using a triangle similarity method. Once an object is identified, the application will provide sound/voice notification, vibration as well as augmented reality image on the screen to inform the user about the obstacle. This is done by having ARCore Software Development Kit. To evaluate the system, tests were carried out which included walking speed limit testing, and usability testing. In the walking speed limit test, the maximum speed when using the application is in the range of 1.27 m/s to 1.35 m/s. Finally, the results of usability testing using Post-Study System Usability Questionnaire (PSSUQ) also show that the application can run well and can be accepted by users with a usability test score of 5.87.