frame_id: /systems/mobile-manipulation-twin
A digital twin for pick-transport-place
An end-to-end mobile manipulation twin — differential-drive AMR plus a 6-DOF cobot — built in Isaac Sim with full ROS 2 control integration.
Problem
A lab-automation platform needed a simulated mobile manipulator capable of autonomous, vision-guided pick-transport-place: navigate to a station, localize a target object, pick it, transport it, and place it — validated in simulation before any physical deployment.
Investigation
Combined an AMR and a 6-DOF cobot into a single URDF-imported articulation in Isaac Sim, exposed full ROS 2 / ros2_control integration for both the base and arm, and tuned Nav2 across five subsystems — controller, costmap, planner, behavior tree, and velocity smoother — to get reliable point-to-point navigation with the arm mounted. Vision-guided picking used an AprilTag/ArUco-based pose estimation service to localize objects at pickup and drop stations.
Outcome
Full autonomous pick-transport-place execution running end to end in simulation, with the same ROS 2 interfaces the physical robot would use — validating the control stack before hardware integration.