MoveIt

MoveIt is an open-source motion planning framework for ROS (Robot Operating System) that enables robots to plan and execute complex movements while avoiding obstacles. Originally developed at Willow Garage and now maintained by PickNik Robotics, it has become the standard software for robotic manipulation research and is used on over 150 robots worldwide.

The framework integrates motion planning algorithms, collision checking, inverse kinematics, and trajectory execution into a unified system. It supports a wide range of industrial and research robots from manufacturers including Universal Robots, KUKA, ABB, Fanuc, Kinova, Franka Emika, and many others.

What it does

MoveIt provides the core capabilities needed for robotic manipulation. The motion planning module generates collision-free trajectories through cluttered environments using sampling-based planners like RRT and PRM. The inverse kinematics solvers compute joint positions for desired end-effector poses, even for over-actuated arms with redundant degrees of freedom.

The collision checking system uses geometric primitives, meshes, or point cloud data to avoid obstacles in real-time. Integration with depth sensors and OctoMaps enables 3D perception for dynamic environments. The manipulation pipeline includes grasp generation libraries for both geometric and machine learning-based approaches.

Integration and deployment

MoveIt 2 runs on ROS 2 Humble, Jazzy, and Rolling distributions. Binary packages are available via apt for Ubuntu 22.04 and 24.04. The framework can also be built from source for custom configurations.

The RViz plugin provides an interactive 3D visualizer for testing planning algorithms before deploying to physical hardware. The Setup Assistant wizard generates configuration packages for new robots. Gazebo integration enables full physics-based simulation for development and testing.

Limitations

• Requires ROS 2 expertise and Linux development environment, creating a steep learning curve for teams new to robotics
• Open-source version lacks the performance optimizations of MoveIt Pro, with significantly slower inverse kinematics and motion planning speeds
• No built-in machine learning capabilities for perception or grasp generation in the open-source version
• Behavior tree orchestration and advanced teleoperation features only available in the commercial MoveIt Pro version
• Limited navigation and whole-body control support compared to integrated commercial platforms
• Hardware abstraction requires manual configuration for each robot model through URDF and SRDF files