Skip to main content

Chapter 2: Isaac ROS 2 Bridge

Learning Objectives​

  • Integrate Isaac Sim with ROS 2 workflows
  • Bridge topics between Isaac Sim and ROS 2
  • Control simulated robots from ROS 2 nodes
  • Use Isaac ROS packages for perception

Introduction​

The Isaac ROS 2 Bridge connects Isaac Sim with ROS 2, enabling:

  • Testing ROS 2 algorithms in Isaac Sim
  • Deploying sim-trained models to real robots
  • Using Isaac ROS perception packages

Enabling the Bridge​

from omni.isaac.core.utils.extensions import enable_extension

# Chapter 2: Enable ROS 2 bridge extension
enable_extension("omni.isaac.ros2_bridge")

Publishing Joint States​

import omni.graph.core as og

# Chapter 2: Create ROS 2 publisher node
keys = og.Controller.Keys
(graph, nodes, _, _) = og.Controller.edit(
    {"graph_path": "/ActionGraph", "evaluator_name": "execution"},
    {
        keys.CREATE_NODES: [
            ("OnPlaybackTick", "omni.graph.action.OnPlaybackTick"),
            ("PublishJointState", "omni.isaac.ros2_bridge.ROS2PublishJointState"),
        ],
        keys.CONNECT: [
            ("OnPlaybackTick.outputs:tick", "PublishJointState.inputs:execIn"),
        ],
        keys.SET_VALUES: [
            ("PublishJointState.inputs:topicName", "/joint_states"),
            ("PublishJointState.inputs:targetPrim", "/World/Robot"),
        ],
    },
)

Subscribing to Commands​

# Chapter 2: Subscribe to /cmd_vel
(graph, nodes, _, _) = og.Controller.edit(
    {"graph_path": "/ActionGraph"},
    {
        keys.CREATE_NODES: [
            ("SubscribeTwist", "omni.isaac.ros2_bridge.ROS2SubscribeTwist"),
            ("DiffController", "omni.isaac.wheeled_robots.DifferentialController"),
        ],
        keys.CONNECT: [
            ("SubscribeTwist.outputs:linearVelocity", "DiffController.inputs:linearVelocity"),
            ("SubscribeTwist.outputs:angularVelocity", "DiffController.inputs:angularVelocity"),
        ],
        keys.SET_VALUES: [
            ("SubscribeTwist.inputs:topicName", "/cmd_vel"),
        ],
    },
)

Isaac ROS Packages​

NVIDIA provides optimized ROS 2 packages:

PackagePurpose
isaac_ros_visual_slamGPU-accelerated SLAM
isaac_ros_image_segmentationReal-time segmentation
isaac_ros_object_detectionYOLO, DOPE
isaac_ros_depth_estimationStereo depth

Example: Visual SLAM​

# Chapter 2: Install Isaac ROS
sudo apt install ros-humble-isaac-ros-visual-slam

# Chapter 2: Launch with Isaac Sim camera
ros2 launch isaac_ros_visual_slam isaac_ros_visual_slam.launch.py

Complete Integration Example​

from isaacsim import SimulationApp
simulation_app = SimulationApp({"headless": False})

from omni.isaac.core import World
from omni.isaac.core.utils.extensions import enable_extension

# Chapter 2: Enable ROS 2
enable_extension("omni.isaac.ros2_bridge")

# Chapter 2: Create world with robot
world = World()
world.scene.add_default_ground_plane()

# Chapter 2: Add robot (URDF or USD)
from omni.isaac.core.utils.stage import add_reference_to_stage
add_reference_to_stage(
    usd_path="/path/to/robot.usd",
    prim_path="/World/Robot"
)

# Chapter 2: Setup ROS 2 publishers/subscribers
# Chapter 2: (Action Graph as shown above)

# Chapter 2: Run simulation
world.reset()
while simulation_app.is_running():
    world.step(render=True)

simulation_app.close()

Key Takeaways​

✅ ROS 2 Bridge connects Isaac Sim to ROS 2 ecosystem
✅ Action Graphs configure topic publishers/subscribers
✅ Isaac ROS provides GPU-accelerated perception
✅ Sim-to-real workflow: train in Isaac, deploy on robot

Previous Section: ← 4.3 Synthetic Data Generation
Next Chapter: Chapter 5: Humanoid Robot Development →