SMART IOT-BASED ROBOT FOR MULTI-PURPOSE SURVEILLANCE APPLICATIONS

Abstract

Surveillance of homes, industrial premises, campuses, warehouses, and border areas is usually carried out with fixed CCTV cameras and human guards. Fixed cameras cover only the area they point at, leaving blind spots, and guards cannot be present everywhere at all times. Certain places, such as tunnels, pipelines, collapsed buildings, and areas with toxic gases, are also dangerous for people to enter. This paper presents a smart IoT-based robot for multi-purpose surveillance that can move through an area, capture live video, sense environmental hazards, and report to a remote operator over the internet. The robot is built on a four-wheel chassis driven by DC motors through a motor driver and controlled by an ESP32 microcontroller. A Raspberry Pi with a camera module handles live video streaming and on-board image analysis. The sensor set includes a passive infrared sensor for detecting human presence, an ultrasonic sensor for obstacle avoidance, a gas sensor for smoke and harmful gases, a flame sensor, and a temperature and humidity sensor. A pan and tilt mechanism allows the camera to look around without moving the whole robot, and infrared lights support observation at night. The robot can operate in two modes. In manual mode, the operator drives it through a web or mobile application while watching the live video feed. In patrol mode, it follows a predefined route, avoids obstacles automatically, and stops to inspect whenever a sensor reports an unusual condition. A lightweight person detection model running on the Raspberry Pi identifies human figures in the camera view, so that an intruder alert is raised only when a person is actually seen, reducing the false alarms that a motion sensor alone would produce. Sensor readings, detections, and snapshots are published to a cloud platform using the MQTT protocol. When an intruder, fire, or gas leak is detected, the system sends push notifications and SMS messages to the owner or security staff together with a captured image and the location of the robot on the patrol map. All events are logged for later review, and recorded video clips can be downloaded from the dashboard.