Lecture Slides

W3L1 - Lecture 5 - Transforms (OM) - Copy.pdf

W3L2 - Lecture 6 - KF & Sensor Fusion.pdf

ROS 1 vs. ROS 2

• ROS 2 has no master node—nodes independently discover each other on the same subnet without needing to start a ROS master.
• Messages can be published with configurable buffer sizes.
• Communication includes QoS (Quality of Service) options for reliability and durability.
• Launch files are now written in Python instead of XML.
• rospy is replaced by rclpy, and roscpp is replaced by rclcpp.
• Specific node base class
• catkin_make is replaced by colcon build
• Some repositories worth checking out
  • https://github.com/DLu/roscompile/blob/main/magical_ros2_conversion_tool/README.md
  • https://github.com/ros2/ros1_bridge

Motivation for Transforms

Until now, we've used sensor data without transformation, assuming the sensor's coordinate frame matched the robot's frame. However, this approach has limitations:

• Different Coordinate Frames
  • Sensors often use their own unique coordinate frames
  • Each sensor may define directions ("forward", "x", "θ") differently
• Transformation Challenges
  • Sensor data needs to be transformed into the robot's frame
  • Multiple sensors increase complexity
  • Combining data from different sensors is challenging

The Solution: A unified robot coordinate frame, typically called "base-link" or "base-footprint", serves as a common reference point for all sensor data.

Transformation Matrix

Sensor positions relative to the base link have 6 degrees of freedom.