ROS 2 for Robot Arms

Learn ROS 2 the way it's used on real arms, in your browser. Write rclpy nodes that command the Pick & Place arm through its controllers, listen to its joint states, read its URDF, publish its tf tree, call and serve services, take parameters, and pick a ball with trajectory and gripper actions. It's real ROS 2 Jazzy Python, ready to move to a real robot.

What you'll learn

• >Write rclpy nodes with publishers, subscriptions, timers and logging
• >Read joint states by name and publish stamped messages in the right frame
• >Parse a URDF and compute forward kinematics from it
• >Broadcast and look up transforms with tf2
• >Call and serve services without blocking the executor
• >Configure nodes with parameters and validate changes
• >Drive ros2_control controllers with trajectory and gripper actions

Before you start

Finish Pick & Place first: this course drives the same arm and reuses its forward kinematics and pick phases. ROS 2 nodes are Python classes, and step 1 explains the little you need. There's nothing to install: a built-in ROS 2 runs in your browser and takes the same rclpy code (Jazzy) you'd run on a real robot. Building packages, launch files and the ros2 command line need a real machine, so the course explains them rather than running them.

Steps

1. 1Your first ROS 2 nodeWrite a node whose timer publishes joint commands to the arm's controller, and make the base wave.Nodes and topicsPublishers and timersLoggingOpen
2. 2Listening to the robotSubscribe to /joint_states, find each joint by name, and publish where the gripper is as a stamped point in base_link.Subscriptions and callbackssensor_msgs/JointStateStamped messagesPro
3. 3Describing the robot (URDF)Read the arm's URDF from its latched topic, compute forward kinematics straight from it, and keep commands inside its joint limits.URDF links and jointsLatched topics (durability)Joint limitsPro
4. 4Frames with tf2Be the robot_state_publisher: broadcast every link's frame, then let tf2 find the gripper and move camera detections into the arm's frame.The tf treeStatic and dynamic transformsLookups with a BufferPro
5. 5Asking and answering (services)Ask the robot where the balls are and how to reach them, visit each one, and offer a /go_home service of your own.ServicesFutures and call_asyncServing requestsPro
6. 6Configuring nodes (parameters)Make the mover's speed and home pose parameters, take them from a launch configuration, and reject bad values at runtime.ParametersOverrides from launch and YAMLValidating changesPro
7. 7Controllers and actionsSwitch the arm to its trajectory controller and pick a ball into the cup with FollowJointTrajectory and GripperCommand goals.ros2_controlActions: goal, feedback, resultJoint trajectoriesPro
8. ★Bonus: A pick-and-place systemSplit pick and place into perception, planning and picking nodes that cooperate over topics, and clear the table.System designAsync callbacksCallback groupsPro