Safety & Health Monitor Node

A periodic check for all nodes to mitigate risk.

---

Potential Tools

• node_monitoring - distinguishes two classes of errors: behavior-level fails & system-level fails. Also provides a way to display the information.
• diagnostic_updater::Heartbeat Class - This seems to be more appropriate for monitoring the actual nodes themselves, in case they were to crash.
• bond - Just like the previous note, it monitors for termination.

Ideas for Manual Implementation

Two topics

• Publish "periodic check" to many nodes (use delays)
• Subscriber responds with "status" (initiate action servers upon bad status)

Actions

• Our Safety & Health Monitoring node would be multiple action clients
• One action server per actuator (services are limited to a single server)
• Suspend functions & react as needed until the issue is resolved
• The servers will raise flags for the client as they provide continuous feedback

Potential Risks & Flags

• Incoming obstacle I(decide between breaking smoothly or initiating the Emergency Stop due to lack of distance).
• Low battery
• Devices/sensors are malfunctioning or need recalibration
• Loss of localization confidence
• High CPU usage