Skip to content

README.md

Latest commit

 

History

History
108 lines (83 loc) · 3.62 KB

README.md

File metadata and controls

108 lines (83 loc) · 3.62 KB

Autonomous Car Robot with Obstacle Avoidance and Infrared Detection (KAYO TEAM)

Overview

This project involves designing, building, and programming a robot capable of both remote-controlled operation and autonomous navigation. The robot Detection lane while avoiding obstacles using Arduino and Raspberry Pi. Part of Kuwait University, Robotics class with Dr. Alshaibani.

Car 3 Car 4

Project Phases:

  1. Phase 1: Assembly and remote operation.
  2. Phase 2: Autonomous driving with obstacle avoidance.

Key Features

  • Remote Control: Manual operation via a web interface (Flask server on Raspberry Pi).
  • Autonomous Navigation: Utilizes Q-learning for path-following and obstacle avoidance.
  • Sensor Integration:
    • 3x IR Sensors for line detection.
    • 3x Ultrasonic Sensors for obstacle detection.
  • Motor Control: L298N motor driver with 4 DC motors.
  • Communication: UART serial between Raspberry Pi (main controller) and Arduino (motor/sensor handler).

Components

Hardware

  • Raspberry Pi 4 (Main controller, runs Flask server).
  • Arduino Uno (Handles motor control & sensor readings).
  • L298N Motor Driver (Controls DC motors).
  • 4x DC Motors with Wheels (Movement).
  • 3x IR Infrared Sensors (Line detection).
  • 3x Ultrasonic Sensors (HCSR04) (Obstacle detection).
  • Power Bank & Li-ion Battery (Power supply).

Software & Tools

  • Python (Flask server, Q-learning algorithm).
  • Arduino IDE (Motor/sensor control logic).
  • GitHub (Version control).

Phase 1: Remote Control

Assembly & Operation

  • Flask web server allows manual control (forward, backward, left, right).
  • Commands sent via UART to Arduino, which drives motors.

Phase 1 Website

Phase 2: Autonomous Mode

Q-Learning Algorithm

  • States: IR sensor readings (e.g., 000 = all white, 010 = middle black).
  • Actions: Forward, left, right, scan_left, scan_right, strongLeft, strongRight.
  • Rewards: Predefined based on sensor inputs (e.g., staying centered = high reward).
  • Training: Robot learns optimal path-following via reinforcement learning.

Obstacle Avoidance

  • Ultrasonic sensors detect obstacles (threshold: 20cm).
  • If an obstacle is detected, the robot switches lanes and signals with LEDs.

Phase 2 Website

Setup & Usage

Setup Guide

Hardware Assembly

  1. Connect motors, sensors, and controllers as per the wiring diagram.
  2. Power Raspberry Pi (USB-C) and Arduino (Li-ion battery).

Software Setup

  1. Upload Arduino code to handle sensors/motors.
  2. Run Flask server on Raspberry Pi (app.py) for web interface.

Training Autonomous Mode

  1. Start training via the web interface.
  2. Robot explores and updates Q-table (model.pth).

Demo

Robot Demo

  • Place robot on a track with lanes and obstacles.
  • Switch between manual and autonomous modes.

Maze

Car 1 Car 2

Challenges & Improvements

Challenges

  • Sensor calibration for varying light conditions.
  • Motor synchronization issues.
  • Q-learning convergence time.

Future Work

  • Fix Issues with Q-learning
  • Add a camera for advanced object recognition.

Contributors

  • Hala Almutairi
  • Zaharaa Alrashidi