Autonomous Control System Simulation Using Image Segmentation

This is the authors' abstract. We don't add key points for this paper.

This paper is dedicated to the problem of image segmentation and the usage of image segmentation in the field of autonomous driving vehicles. It provides an analysis of different systems used for image segmentation and compares multiple methods for providing information about the distances of different objects in space. More specifically, it looks at the YOLACT system which is then implemented into the mobile application. Using ultrasonic sensors connected to the Arduino platform, it can measure distances from the vehicle model, on which sensors are placed together with the mobile phone to scan the surroundings. The proposed methodology combines real-time image segmentation with sensor-based distance measurement using custom software implemented on mobile and Arduino platforms. The final data provided by image segmentation are combined with distance information to simulate an autonomous driving system in the form of commands for a vehicle. The final solution is tested and the results are evaluated. In evaluation, the usage of the specific technologies is evaluated as well as the fulfilment of requirements for the solution.