SENTINEL: Autonomous Emergency Ground Intelligence System for Rescue
An autonomous ground search and rescue robot integrating YOLOv8 human detection, dynamic radar navigation, telemetry sensing, and a 3-DOF robotic manipulator.
Overview
Search and rescue operations in structural collapse, fire, and disaster environments require ground robotic systems capable of navigating unpredictable terrain while locating trapped individuals and transmitting real time environmental data. SENTINEL is a ground intelligence platform built around a multi controller hardware interfacing architecture to handle low level motor control, dynamic obstacle avoidance, vision processing, and wireless telemetry streaming. The system utilizes a Master Slave hardware topology. An Arduino Uno serves as the master real time controller, responsible for motor drive execution, sensor data acquisition, pulse width modulation signal generation, and servo arm kinematics. An ESP8266 NodeMCU or ESP32 S3 acts as the communication slave, hosting an onboard WiFi web dashboard, managing network API endpoints, and processing incoming telemetry packets. Communication between the controllers is mediated over a bidirectional UART serial bus with resistive voltage level shifting to bridge 5 volt and 3.3 volt logic levels safely. Autonomous navigation is governed by a dynamic proportional speed control algorithm paired with a servo mounted ultrasonic radar scanner. An HC SR04 ultrasonic sensor mounted on an SG90 servo continuously sweeps a 180 degree frontal sector. Rather than executing rigid grid turns, the master controller calculates proportional PWM motor speeds based on directional distance clearances, allowing the vehicle to execute smooth continuous curved paths around obstacles. If trapped in a corner or dead end, an integrated recovery state machine detects repeated directional changes, executes a timed reverse maneuver, and re scans the environment for open pathways. Human detection is executed through an edge vision pipeline. A camera feed is processed in real time using YOLOv8 and OpenCV. When a person is detected with high confidence, an HTTP POST trigger is dispatched over WiFi to the ESP web server. The communication slave immediately relays a high priority stop command over the UART bus to the master controller, halting vehicle movement, sounding a local buzzer alarm, and logging the exact NEO 6M GPS coordinates on the web dashboard. Environmental hazard monitoring is maintained via a DHT11 sensor reporting real time ambient temperature and humidity metrics to detect thermal risks. For physical intervention, SENTINEL features a 3-DOF robotic arm powered by three SG90 servo motors, providing base rotation, shoulder elevation, and gripper claw control to clear light debris or assist survivors during rescue missions.
Tech Stack
Empirical Benchmarks
Master Slave Topology (Arduino Uno Master & ESP Slave)
Microcontroller Architecture
YOLOv8 Person Class via OpenCV Python Pipeline
Object Detection Model
Dynamic Proportional PWM (20cm to 60cm thresholding)
Obstacle Avoidance Radius
180 degree frontal coverage via SG90 Servo Mount
Radar Scanner Sweep
3-DOF Servo Arm (Base, Shoulder, Gripper Claw)
Manipulator Degrees of Freedom
NEO 6M GPS, DHT11 Temperature Humidity, HC SR04 Ultrasonic
Telemetry Sensors