/Automated-Construction-Site-Inspector

Developed an IoT-based construction site inspector using a Raspberry Pi 4 that autonomously navigates and inspects construction sites. The system features two DC motors for line-following and a servo-mounted ultrasonic sensor for real-time obstacle detection.

Primary LanguagePython

Real-Time Robot Control System

Watch the Video

Watch the video on Google Drive

Overview

This project demonstrates a real-time robot control system using Raspberry Pi and various sensors, motors, and cameras. It is designed to navigate autonomously, avoid obstacles, track black lines, and stream real-time video for AI-based analysis using YOLO and GPT-4V. The project was developed by Haim Ozer and Shon Pazarker to showcase skills in real-time embedded systems, robotics, and AI integration.

Features

  • Motor Control: The robot uses two DC motors controlled via an H-Bridge module for forward, backward, and directional movements.
  • Servo Motors: Three servo motors control the ultrasonic sensor for obstacle avoidance and a camera for panoramic view (horizontal and vertical movements).
  • Camera Module: A USB camera streams real-time video which is uploaded via Wi-Fi to a remote server for AI processing.
  • Obstacle Avoidance: An ultrasonic sensor mounted on a servo detects obstacles and adjusts the robot's path.
  • Line Tracking: Four infrared sensors enable the robot to follow a black line on the ground.
  • AI Integration: Video streams are processed by a remote server using YOLO and GPT-4V for object detection and analysis. A PDF report is generated with a frame-by-frame timeline of detected objects.

Hardware Components

  1. 3 Servo Motors:
    • 1 for obstacle avoidance (ultrasonic sensor rotation).
    • 2 for the camera's horizontal and vertical movement.
  2. 2 DC Motors: Controlled via H-Bridge for movement.
  3. USB Camera: Streams video for remote AI processing.
  4. Ultrasonic Sensor: Mounted on a servo motor for real-time obstacle detection.
  5. 4 Infrared Line Tracking Sensors: Constantly monitor the robot's position on a black line.

Software Components

  • Real-time Motor Control: The robot uses GPIO pins to control the motors, ensuring precise movement based on sensor input.
  • Video Streaming: A threading mechanism captures video from the camera and uploads it via Wi-Fi to a remote server.
  • AI Video Processing: YOLO and GPT-4V analyze the uploaded video, generating a PDF report summarizing detected objects with a timeline.
  • Obstacle Avoidance Algorithm: The ultrasonic sensor rotates to scan for obstacles and adjust the robot’s direction in real-time.
  • Line Tracking Algorithm: The robot stays on track by constantly adjusting its direction based on input from the infrared sensors.

System Architecture

  • Real-time Control Loop: The robot constantly processes sensor data and adjusts its movement based on the detected environment, ensuring smooth navigation.
  • Wi-Fi Video Upload: Real-time video is captured and uploaded to a cloud server for AI analysis.
  • Remote Processing: The video is analyzed using YOLO for object detection and GPT-4V for frame-by-frame commentary, generating a detailed PDF report.
  • PID Control for Motors: Ensures accurate motor control, maintaining the robot's speed and direction in response to sensor feedback.

Key Technologies

  • Raspberry Pi GPIO for Motor and Sensor Control
  • Real-time Multithreading for Video Capture and Upload
  • OpenCV for Video Processing
  • Wi-Fi Communication for Video Upload
  • YOLO & GPT-4V for AI Processing