Skip to main content

2.7 Exercises and Summary

Hands-On Exercises​

Exercise 1: Test Gravity Toggle​

Goal: Spawn robot in Gazebo, toggle gravity, observe behavior

Steps:

  1. Launch Gazebo: ros2 launch digital_twin_chapter2 gazebo_sim.launch.py
  2. Run gravity test: python3 test_gravity.py
  3. Observe robot floating vs. falling

Expected Outcome: Robot floats when gravity=0, falls at 9.81 m/s² when gravity=-9.81

Validation Criteria:

  • ✅ Robot spawns without errors
  • ✅ Gravity toggle changes robot motion
  • ✅ No inter-penetration with ground plane

Exercise 2: Create Unity Scene with Human​

Goal: Import URDF to Unity, add human character, test HRI reachability

Steps:

  1. Open Unity, create new scene
  2. Import URDF: Assets → Import Robot from URDF
  3. Add human character from Unity Asset Store
  4. Position human 1m in front of robot
  5. Test arm reach

Expected Outcome: Robot arms can reach human within 1.5m radius

Exercise 3: Simulate LiDAR and Visualize​

Goal: Add LiDAR to URDF, launch Gazebo, visualize in RViz

Steps:

  1. Add LiDAR sensor block to URDF
  2. Launch Gazebo
  3. Launch RViz: rviz2
  4. Add LaserScan display, topic: /scan

Expected Outcome: Red scan points showing obstacles

Exercise 4: Run Complete Digital Twin​

Goal: Launch entire pipeline with one command

Steps:

  1. ros2 launch digital_twin_chapter2 digital_twin_complete.launch.py
  2. Open Unity, press Play
  3. Verify robot moves in both Gazebo and Unity

Expected Outcome: Perfect sync between Gazebo physics and Unity rendering

Chapter Summary​

Key Concepts Learned​

  1. Digital Twins: Virtual replicas enable safe testing, rapid iteration
  2. Gazebo Physics: Gravity, friction, collision detection, joint dynamics
  3. ros2_control: Hardware-agnostic controller interfaces
  4. Unity Rendering: Photorealistic visualization for demos and HRI
  5. Sensor Simulation: LiDAR, cameras, IMUs with realistic noise
  6. Integration: Single-command launch orchestrates entire pipeline

Technical Skills Acquired​

  • ✅ Install and configure Gazebo Classic for ROS 2
  • ✅ Create physics-accurate URDF models with inertial properties
  • ✅ Configure ros2_control for position/velocity/effort control
  • ✅ Import URDFs to Unity with ROS-Unity bridge
  • ✅ Simulate sensors (LiDAR, depth camera, IMU) with noise
  • ✅ Validate simulation accuracy (gravity, collision, joint limits)
  • ✅ Orchestrate multi-component systems with hierarchical launch files

Common Pitfalls Avoided​

  • ❌ Missing <inertial> tags → Robot explodes in Gazebo
  • ❌ Collision meshes too complex → Real-time factor < 0.5
  • ❌ Forgetting use_sim_time: True → TF tree desynchronized
  • ❌ Incorrect joint limits → Controller rejects commands
  • ❌ No sensor noise → Sim-to-real gap causes hardware failures

Looking Ahead: Module 3​

Next Module (Chapters 8-10) transitions to:

  • NVIDIA Isaac Sim: GPU-accelerated physics (100x faster)
  • Photorealistic rendering: RTX ray tracing for camera simulation
  • VSLAM & Nav2: Visual SLAM and autonomous navigation
  • Isaac ROS: Hardware-accelerated perception pipelines

What Transfers Directly:

  • ✅ URDF models (Isaac Sim imports URDF natively)
  • ✅ ros2_control architecture
  • ✅ Sensor configuration patterns
  • ✅ Launch file structure

What Changes:

  • World files: Gazebo .world → Isaac USD (Universal Scene Description)
  • Physics: ODE (CPU) → PhysX 5 (GPU)
  • Rendering: Ogre → RTX ray tracing

Additional Resources​

Congratulations!​

You now have a complete digital twin pipeline that enables:

  • Safe algorithm testing before hardware deployment
  • Rapid iteration (minutes vs. weeks)
  • Realistic sensor simulation with noise models
  • High-fidelity visualization for stakeholders

This is the foundation for all advanced Physical AI development.

Ready for Module 3? Proceed to Chapter 8: NVIDIA Isaac Sim and GPU-Accelerated Simulation