Fault injection scenarios for Turtlebot diagnostics demo

[bburda]

Summary

Add fault injection scenarios to the TurtleBot3 demo that showcase how ros2_medkit enables diagnosing and managing robot problems through the REST API and Web UI.

Proposed solution

Goals

1. Demonstrate fault detection - Show faults appearing in Web UI
2. Demonstrate fault analysis - View fault snapshots with context
3. Demonstrate recovery - Clear faults and observe system healing
4. Realistic scenarios - Real robotic failure modes

Proposed Fault Scenarios

Scenario 1: Sensor Timeout (LiDAR)

Simulated LiDAR driver stops publishing → detected as sensor fault

User flow:

1. Robot navigating normally
2. Click "Inject LiDAR Fault" button (via service call)
3. See LIDAR_TIMEOUT fault appear in Web UI Faults panel
4. Robot stops moving (Nav2 detects no scan data)
5. Click "Clear Fault" → Fault clears, LiDAR resumes

Scenario 2: Motor Overheat

Simulated motor temperature exceeds threshold:

Fault memory snapshot shows temperature at fault time:

{
  "motor_temp": 85.0,
  "ambient_temp": 25.0,
  "duty_cycle": 0.95,
  "runtime_hours": 2.3
}

Scenario 3: Localization Lost

AMCL reports high uncertainty → localization fault:

User flow:

1. Robot navigating
2. Teleport robot via Gazebo (breaks localization)
3. See LOCALIZATION_LOST fault
4. Re-localize via Initial Pose → fault auto-clears

Scenario 4: Navigation Stuck

Robot unable to find path → planner fault

Scenario 5: Robot Collision (Multi-Robot)

Two TurtleBots collide while navigating → fleet-level fault

User flow:

1. Send robot1 and robot2 to crossing paths
2. Both robots detect collision (bumper contact in Gazebo)
3. See ROBOT_COLLISION faults on both robots in Web UI
4. Faults are correlated by timestamp and position
5. Fleet manager can coordinate recovery (backup both robots)
6. Clear faults → robots resume with new paths