INTEGRATION OF HUMAN WITH IMAGE-BASED MODELS, AND ROBOTICS FUNDAMENTALS USING WEARABLE IMMERSIVE PLATFORMS

This dissertation enhances human-computer interactions through integration of pattern recognition, Deep Learning (DL), and robotics fundamentals with C#-UnityEngine platform Augmented Reality (AR) headsets. This document begins by introducing a method to integrate Canny algorithm with AR headsets’ platform for crack detection. It then develops a method for converting pixels to engineering scales for crack measurements. Additionally, to reduce the runtime of image processing, the research proposes an automatic Region-Of-Interest (ROI) selection algorithm. Next, the integration of DLs with AR headsets allows for complex image recognition tasks. This thesis demonstrates the value of integrating human awareness of the environment with robotic tasks by creating AR interfaces for robot programming that allow users to reduce randomness of robots’ sampling-based path-planning. The ultimate goal of this research is to evaluate enhancement of human-machine collaboration in robotics and visual inspection by creating the mentioned AR platforms and integrations.