Autonomous Ping Pong Ball Collector

This paper aims to document and showcase the process of a final-year Computer Engineering Capstone project at Toronto Metropolitan University. The goal was to create a fully autonomous vehicle capable of detecting and collecting scattered ping pong balls across a flat plane through the use of computer vision, machine learning and computer engineering. This involved several key components such as collecting an image dataset through a high res camera, training a machine learning model on said dataset using the Mobilenet SSD V2 neural network architecture for the purposes of object detection, assembling and integrating hardware components such as motors and servos with their respective wirings and power management, working with NVIDIA Jetson Nano and it's interfaces, and using python/C++ to program scripts to help coordinate all of the different functions. This project was a collaborative effort that involved the integration of these different software and hardware components, and this paper attempts to concisely highlight the technical details of each component and explain how it integrates with the rest of the system in order to achieve the goal of autonomously collecting ping pong balls. Upon completion of the capstone term, the project was deemed a success and was demonstrated to the respective FLC, and was shown at the capstone showcase at Toronto Metropolitan University.