🛡️Sensor Guard (ROS 2)

A real-time safety layer for robotic velocity commands, designed to handle noisy inputs, unstable control signals, and system-level failures.

Built in ROS 2, this node enforces safe motion constraints, anomaly detection, and fault recovery for mobile robots.

---

🔷 System Overview

/cmd_vel_raw → [Sensor Guard] → /cmd_vel
│
├── /safety_state
└── /sensor_guard/reset_fault

---

Core Capabilities

🧮 1. Signal Conditioning

• Deadband filtering to suppress noise
• Hard limits on linear and angular velocity

📊 2. Statistical Robustness

• Sliding-window filtering
• Outlier rejection using mean + standard deviation
• Rate limiting for smooth motion (bounded acceleration)

🧠 3. System Awareness

• Detects command stream failure (timeout-based)
• Publishes real-time system state:
  • NORMAL
  • DEGRADED
  • FAULT

📉 4. Quality Monitoring

• Tracks how often commands are corrected
• Automatically flags degraded input quality

🚨 5. Fault Handling (Latched Safety)

• Fault state persists until explicitly reset
• Robot is forced to stop during FAULT

Reset service:

ros2 service call /sensor_guard/reset_fault std_srvs/srv/Trigger {}

---

How to Run

colcon build --symlink-install
source install/setup.bash
ros2 run sensor_guard cmd_vel_guard

Test Input

ros2 topic pub -r 10 /cmd_vel_raw geometry_msgs/msg/Twist "{linear: {x: 0.5}, angular: {z: 1.0}}"

Observe Outputs

ros2 topic echo /cmd_vel
ros2 topic echo /safety_state

---

Fault Injection

Stop publishing /cmd_vel_raw

→ System transitions to FAULT
→ Robot output forced to zero

Reset:

ros2 service call /sensor_guard/reset_fault std_srvs/srv/Trigger {}

---

Future Extensions

• ROS 2 diagnostics integration (diagnostic_msgs)
• Visualization dashboards (Foxglove / rqt)
• Metrics tracking (latency, correction rate)
• Integration with Nav2 / autonomy stack