Advanced Autonomous Cars Design Using ROS (Level 4)

Advanced Autonomous Cars Design Using ROS (Level 4)

Target Audience: Undergraduate Mechatronics Students
Prerequisite: Completed Basic ROS Course
Total Duration: 50 hours (10 lectures × 5 hours)

Course Objectives
By the end of this course, students will:
• Integrate advanced perception and control modules using ROS
• Implement path planning, SLAM, object tracking, and sensor fusion
• Simulate and test autonomous behaviors in Gazebo or real-time platforms
• Complete a final integrated project showcasing end-to-end autonomous car logic

Course Outline:
Lecture 1: ROS Navigation Stack and Cost maps
Theory
• Overview of the ROS Navigation Stack (move base)
• Structure of cost maps: global vs. local
• Understanding inflation layers and obstacle layers
• Role of TF, odometry, and sensor input in navigation
Practical
• Configure and launch the ROS navigation stack in simulation
• Tune cost map parameters (inflation radius, obstacle layer thresholds)
• Visualize cost maps and trajectory paths in RViz
• Perform basic autonomous navigation in a static map

Lecture 2: Localization Using AMCL and Extended Kalman Filters (EKF)
Theory
• Principles of robot localization
• Monte Carlo Localization (AMCL) explained
• Extended Kalman Filter (EKF) for sensor fusion
• TF trees and coordinate frame transformations
Practical
• Set up AMCL package and test localization using LiDAR and odometry
• Configure robot_pose_ekf or robot_localization package
• Fuse IMU, GPS, and odometry using EKF
• Monitor localization accuracy in RViz

Lecture 3: SLAM – Simultaneous Localization and Mapping
Theory
• What is SLAM? Types of SLAM: Gmapping, Hector, Cartographer
• Differences between SLAM and localization
• Use cases in urban autonomous navigation
Practical
• Launch SLAM packages in Gazebo or real robot
• Build a live map while navigating a simulated environment
• Save and reload map using map_server
• Evaluate SLAM performance with and without obstacles

Lecture 4: Path Planning and Obstacle Avoidance
Theory
• Fundamentals of path planning (global vs. local)
• Common algorithms: A*, Dijkstra, DWA (Dynamic Window Approach)
• Dynamic obstacle handling in real environments
Practical
• Implement global path planning using move_base and static map
• Test local planning using DWA and clear cost maps
• Add dynamic obstacles in Gazebo and observe path re-planning
• Compare different planners and tune parameters

Lecture 5: Sensor Fusion (LiDAR + Camera + IMU)
Theory
• Concept of sensor fusion: advantages and challenges
• Synchronization of asynchronous sensors
• Kalman Filters and data association basics
Practical
• Collect and process data from LiDAR, camera, and IMU
• Use robot_localization or custom EKF to fuse sensor data
• Visualize sensor fusion results in RViz
• Create TF trees for correct sensor alignment

Lecture 6: Vision-Based Lane Detection and Object Tracking
Theory
• OpenCV techniques for lane detection
• Edge detection, color thresholding, and perspective transforms
• Object detection techniques using YOLO or SSD
Practical
• Build a lane detection pipeline using OpenCV
• Integrate camera data with ROS node
• Apply a pre-trained YOLO model for real-time object detection
• Publish object positions to ROS topics

Lecture 7: Vehicle Control – Kinematic Models and PID Tuning
Theory
• Differential vs. Ackermann kinematics
• Motion constraints and turning radius
• Introduction to PID control and tuning techniques
Practical
• Apply a kinematic model in Gazebo
• Design a PID controller for steering and throttle
• Tune PID parameters for smooth tracking
• Test vehicle response in different map conditions

Lecture 8: Behavior Trees and Decision-Making
Theory
• Reactive vs. deliberative systems
• Introduction to Behavior Trees (BT) and state machines
• Use cases in robotic decision-making
Practical
• Install and configure ROS BehaviorTree.CPP
• Design a basic BT for navigation, stop, avoid, and resume behaviors
• Simulate different decision paths based on sensor input
• Integrate BT with navigation stack

Lecture 9: Advanced Simulation in Gazebo
Theory
• Building custom environments in Gazebo
• Creating accurate URDF/SDF models of autonomous cars
• Adding traffic signs, lights, pedestrians for realism
Practical
• Import or design a custom city simulation environment
• Configure robot model with sensors, controllers, and actuators
• Simulate real-world traffic conditions and dynamic elements
• Run complete navigation and object detection scenarios

Lecture 10: Final Integrated Project
Theory
• Explanation of the final project: Urban Navigation with Real-Time Obstacle Avoidance and Object
Detection
• Integration plan: sensors, control, vision, planning
• Evaluation criteria and demonstration format
Practical
• Teams implement a fully integrated system using ROS
• Demonstrate map loading, AMCL localization, obstacle avoidance, lane tracking, and decision logic
• Record and present the system workflow and performance
• Each student presents their role and understanding

if you would like to get our course content please register . . .