Drishti - An Autonomous Surveillance Bot Using Raspberry Pi

Abstract

This paper introduces Drishti, an intelligent autonomous monitoring robot built on the Raspberry Pi 4 platform to provide productive and effective, real-time observation. The system combines several sensor modules; vision, ultrasonic distance sensing, inertial measurement, and GPS/GSM communication; to support self-navigation, collision avoidance, and independent decision-making. By integrating computer vision techniques with lightweight deep learning algorithms, Drishti can identify items, recognize faces, and evaluate potential threats within its surroundings. A hybrid SLAM architecture, including DROID-SLAM, SuperGlue, and a Neo4j-based Graph Retrieval-Augmented Generation (GraphRAG) network, enables precise mapping and context-aware localization. The robot moreover features a custom battery management and encryption system that enhances energy productivity and data security. Experimental validation shows that Drishti delivers consistent and stable operation, low-latency wireless communication, and stable performance in dynamic environments The proposed design provides a compact, low-cost, and scalable approach for intelligent monitoring across residential, industrial, and public-safety domains.