Module 2: The Digital Twin

In Module 1, we built the "Nervous System." Now, we build the world. A Digital Twin is more than just a 3D model; it is a virtual mirror of physical reality, complete with gravity, friction, and sensor noise.

Learning Objectives

• Understand the Sim-to-Real pipeline.
• Learn the differences between URDF and SDF.
• Configure physics engines for stable humanoid locomotion.
• Simulate LiDAR and Depth Cameras for perception training.
• Evaluate the use of Unity for human-robot interaction.

---

1. The Physics of Reality

For an AI to learn how to walk, the simulation must respect the laws of physics. ROS 2 primarily uses Gazebo for this purpose.

Physics Engines

• ODE (Open Dynamics Engine): The default for Gazebo. Fast and stable for rigid body dynamics.
• Bullet: Known for its advanced collision detection and soft body dynamics.
• DART: Excellent for articulated robots like humanoids.

:::info Pro Tip When simulating humanoids, high-frequency physics updates (e.g., 1000Hz) are often necessary to maintain balance. :::

---

2. Gazebo: The Robot's Sandbox

Gazebo uses SDF (Simulation Description Format), an XML format that extends URDF to include simulation-specific properties like friction, surface materials, and sensor plugins.

Adding a LiDAR Sensor

To give your robot eyes, you must add a sensor plugin to your SDF file:

<sensor name="lidar" type="gpu_ray">
  <pose>0 0 0.5 0 0 0</pose>
  <update_rate>10</update_rate>
  <ray>
    <scan>
      <horizontal>
        <samples>640</samples>
        <resolution>1</resolution>
        <min_angle>-1.57</min_angle>
        <max_angle>1.57</max_angle>
      </horizontal>
    </scan>
    <range>
      <min>0.1</min>
      <max>30.0</max>
    </range>
  </ray>
  <plugin name="gazebo_ros_ray_sensor" filename="libgazebo_ros_ray_sensor.so">
    <ros>
      <remapping>~/out:=scan</remapping>
    </ros>
    <output_type>sensor_msgs/LaserScan</output_type>
  </plugin>
</sensor>

---

3. Unity: High-Fidelity Interaction

While Gazebo excels at physics, Unity is the leader in visual fidelity and complex environmental interactions.

Why Use Unity?

1. Synthetic Data: Generate photorealistic images for training computer vision models.
2. Human-Robot Interaction (HRI): Simulate humans in the environment with high-quality animations.
3. VR/AR Integration: Test how humans might interact with your robot using a VR headset.

---

4. SDF vs. URDF: A Comparison

Feature	URDF	SDF
Primary Use	ROS 2 State/Kinematics	Gazebo Simulation
World Support	No (Single Robot only)	Yes (Multiple Robots + Worlds)
Friction/Contact	Limited	Comprehensive
Complexity	Simple	High

---

Summary

By creating a Digital Twin, you have built a risk-free playground for your AI. In the next module, we will introduce the AI-Robot Brain using the NVIDIA Isaac platform, where simulation meets hardware acceleration.

:::caution Hardware Check Running Gazebo or Unity requires an NVIDIA RTX GPU for real-time performance. Ensure your drivers are updated! :::