Self-Driving Cars with Duckietown

This course provides a hands-on introduction to the key concepts and technologies behind self-driving cars using the Duckietown platform. Learners will explore autonomous navigation, computer vision, control systems, and machine learning, all within the context of Duckietown’s miniaturized self-driving car environment.

---

Module 1: Introduction to Self-Driving Cars and Duckietown 

Overview of Autonomous Vehicles:

History and future of self-driving cars
The five levels of autonomy in self-driving systems

Overview of the Duckietown educational platform
Hardware setup: Duckiebot assembly and components (camera, motors, sensors)
Software overview: Duckietown operating system, ROS (Robot Operating System) basics

---

Module 2: Robot Kinematics and Motion Control

Introduction to Robot Kinematics:

Differential drive kinematics and odometry
Robot motion and steering

Proportional-integral-derivative (PID) control for autonomous motion
Controlling speed and direction of Duckiebots

---

Module 3: Computer Vision for Autonomous Driving

Image Processing Fundamentals:

Introduction to computer vision and OpenCV
Pre-processing techniques: Thresholding, edge detection, and contour detection

How to detect road lanes using computer vision
Applying the Hough Transform for line detection
Identifying lane markings and keeping the Duckiebot on track

Techniques for detecting and tracking objects (e.g., other vehicles, pedestrians)
Using bounding boxes and color segmentation for obstacle avoidance

---

Module 4: Localization and Mapping

Introduction to Localization:

Concept of localization and why it is important for self-driving cars
Understanding sensor fusion techniques (e.g., combining camera data with odometry)

Simultaneous Localization and Mapping (SLAM):Basics of SLAM for building maps while navigating the environment
How Duckiebot uses SLAM to create and update maps of the Duckietown environment

---

Module 5: Path Planning and Obstacle Avoidance (10 Hours)

Path Planning Algorithms:

Overview of different path planning algorithms (A*, Dijkstra, RRT)
How to plan and execute paths for autonomous navigation in a dynamic environment

Real-time obstacle detection and avoidance techniques
Reactive control methods: Handling dynamic obstacles such as moving cars or pedestrians

---

Module 6: Deep Learning for Autonomous Driving (5 Hours)

Introduction to Deep Learning for Self-Driving Cars:

Overview of neural networks and their application in autonomous vehicles
Using Convolutional Neural Networks (CNNs) for object recognition and lane detection

Collecting training data using Duckiebot’s camera
Training models for visual navigation and obstacle detection

Introduction to vehicle-to-infrastructure communication
How Duckiebot communicates with traffic lights, stop signs, and other infrastructure

Detecting and responding to traffic lights and stop signs using computer vision
Implementing traffic management rules in Duckietown (e.g., stop at red lights)

Program Duckiebot to recognize traffic signals and manage traffic interactions

---

Course Prerequisites:

Basic programming knowledge in Python or a similar language
Familiarity with Linux and command-line interfaces is helpful but not required
Basic understanding of robotics, sensors, and control systems is beneficial but not mandatory

---