Your cart is empty.
Keep shoppingROS 2 Robotic Arm Mastery: MoveIt2, Gazebo & VLA Models
New
29 students
ROS 2 Robotic Arm Mastery: MoveIt2, Gazebo & VLA Models
Build, simulate and control a robot arm with ROS 2, URDF, MoveIt2 and Gazebo, from first joint to pick-and-place
Created byFerbin Richard
Last updated 8/2026
English
English
What you'll learn
- Build a complete robot arm description from scratch in URDF and Xacro, with links, joints, inertias, collision geometry and ros2_control interfaces.
- Plan and execute collision-aware arm motions with MoveIt2, using the Setup Assistant, OMPL planners, RViz MotionPlanning, and joint vs Cartesian goals.
- Simulate a manipulator in Gazebo with real physics: spawn the robot, configure controllers, command joints from ROS 2, and debug spawn failures.
- Design and attach grippers, reason about force closure and grasp planning, and use attach/detach so the planner treats a grasped object correctly.
Explore related topics
Course content
13 sections • 116 lectures • 3h 41m total length
Requirements
- Basic Python familiarity is enough. No prior ROS 2, MoveIt or robotics experience needed, everything is built up from first principles.
- A computer running Ubuntu 24.04 with ROS 2 Jazzy, or a virtual machine. Section 2 walks through the full install step by step.
- No robot hardware required. Every lecture runs in simulation with Gazebo, so a laptop is all you need to follow along.
- Comfort with a Linux terminal helps but is not assumed. Basic commands are covered before you need them.
Description
This course contains the use of artificial intelligence.
Learn to build, simulate and control a robotic manipulator from the ground up, using ROS 2 and
MoveIt2.
Most robotics courses stop at mobile robots. This one goes where the jobs are: manipulation.
You will build a complete robot arm from an empty file, describe it in URDF and Xacro, give it
physics and controllers, plan collision-free motions with MoveIt2, attach a gripper, and
finish with a full pick-and-place system you can put in a portfolio.
Everything runs in simulation. No robot hardware required.
You start with the fundamentals: links, joints, degrees of freedom, and the maths that
actually matters for manipulation. Transforms and kinematics are derived on screen, not
asserted on a slide. Then you build: URDF and Xacro, meshes and collision geometry,
ros2_control, and spawning your arm into Gazebo with real physics underneath it.
From there you go deep on MoveIt2 across a full section: the Setup Assistant, OMPL planners,
the planning scene, joint versus Cartesian goals, and RViz MotionPlanning. You will also cause
failures deliberately and fix them on camera, because reading an error message is the skill
that actually transfers.
What sets this course apart: a full section on world models and Vision-Language-Action
policies, covering where learned methods genuinely help, where classical planning still wins,
and how the two fit together. Very few manipulator courses touch this at all.
You will also cover grippers and grasp mechanics, force closure, attach and detach so the
planner treats a grasped object correctly, camera and depth perception feeding into
manipulation, reinforcement learning foundations, and an honest look at the sim-to-real gap.
By the end you will have built a working manipulation stack end to end, and understand every
layer well enough to debug it when it breaks.
Who this course is for:
- Robotics beginners who know some Python and want to move from theory into building and controlling a real robot arm in simulation.
- Engineering and CS students who need practical, portfolio-ready manipulator experience that coursework and textbooks rarely provide.
- Career changers targeting robotics roles who want an end-to-end project, from URDF to pick-and-place, they can show in interviews.