ARYA: An AI-Powered Campus Assistant Robot Integrating Real-Time Computer Vision, Conversational AI, and 3D DigitalTwin Simulation

Abstract: Campus environments increasingly require interactive systems that can perceive surroundings, recognize individuals, converse naturally, and communicate robotic state. This paper presents ARYA, an AI-powered campus assistant robot platform integrating real-time computer vision, face recognition, conversational AI, and 3D digital-twin simulation. A Python/FastAPI backend runs YOLOv8-based object and person detection with an InsightFace recognition pipeline, while a React, TypeScript, and Three.js frontend provides a real-time robot avatar and browsable floor-plan digital twin. The conversational subsystem combines rule-based campus-information handling with a Gemini-based generative-AI fallback. Voice and typed commands control the simulated avatar and, where available, GPIO-based hardware. The paper describes the system architecture, perception and greeting pipelines, REST, WebSocket and MJPEG

be developed and evaluated in simulation before physical deployment.

Keywords: service robot, human–robot interaction, computer vision, YOLOv8, face recognition, InsightFace, conversational AI, digital twin, FastAPI, React Three Fiber.