This thesis presented the design and implementation of an IoT-controlled robotic system that integrates machine vision and virtual assistant technologies to enable intuitive and context-aware human-robot interaction. Motivated by the need for more accessible and intelli-gent robotic systems in domestic and assistive contexts, the project explored how natural language interfaces, real-time visual perception, and modular robotic control could be unified to perform complex manipulation tasks in constrained environments.The developed system integrated an open-source robotic arm with a cloud-based voice interface using Amazon Alexa and a real-time machine vision module powered by YOLOv8 and OpenCV.

The Robot Operating System (ROS) served as the middleware platform, man-aging communication between the voice interface, perception system, and actuation mod-ules. This architecture enabled the robot to receive spoken commands, detect and localize objects within the scene, and execute corresponding manipulation tasks. The implementation demonstrated successful task execution, confirming the system’s ability to link voice-based interaction with reliable visual perception in a modular, low-cost configuration.The research addressed two primary questions related to the system’s technical viability and its impact on usability.

Findings confirmed that virtual assistant integration significantly enhances accessibility while maintaining operational precision. Despite limitations such as reliance on monocular vision and open-loop control, the prototype proved effective in per-forming real-time object-based tasks. The thesis concludes by outlining avenues for future work, including integration of depth sensing, on-device AI, dialogue-based interaction, and adaptive learning. Collectively, this work contributes to advancing intelligent, voice-enabled robotic systems designed for human-centric environments.