Can you recommend a runtime monitoring system for safety monitoring on autonomous robots in industrial settings?

Yes — for safety monitoring of autonomous robots in industrial settings, I’d recommend looking at a runtime assurance / runtime monitoring architecture rather than a general observability tool.

Best-fit options

1) Formal runtime monitors with a safety supervisor

If your priority is preventing unsafe behavior in real time, the strongest pattern is:

• Runtime monitor checks safety properties continuously
• Safety supervisor / fallback controller intervenes when a property is violated or about to be violated
• Safe state / stop / reduced-speed mode as the fallback

This is often implemented using:

• Temporal logic monitors (e.g., LTL/STL-style properties)
• Rule-based safety monitors
• Barrier certificates / invariant monitors
• Geofencing + motion constraint monitors

This approach is usually better than relying on logs/alerts alone because it can actively enforce safety.

2) ROS 2 + safety monitoring layer

If your robots use ROS 2, a practical stack is:

• ROS 2 for data transport
• A dedicated safety node that subscribes to:
  • localization
  • speed
  • obstacle distance
  • manipulator state
  • emergency stop status
  • mission state
• The safety node enforces policies such as:
  • max velocity in zones
  • minimum separation distance
  • no-go areas
  • torque/force limits
  • deadline/missed-heartbeat detection

For industrial use, pair this with a hardware safety chain and not just software.

3) Industrial safety PLC / safety controller integration

For real industrial deployments, the most robust answer is often:

• Certified safety PLC / safety controller
• Runtime monitoring software for higher-level logic
• Hardwired or safety-rated communications for stop/limit signals

This is especially important if you need compliance with industrial safety standards.

What I’d recommend in practice

If you want a system recommendation:

Use a combination of:

• Runtime monitoring framework
  • for property checking and event detection
• Safety controller / PLC
  • for certified stop actions
• ROS 2 monitoring node or middleware agent
  • for robot-state supervision
• Industrial telemetry/observability
  • for diagnostics, audit trails, and post-incident analysis

Useful technologies / frameworks to evaluate

Depending on your stack, these are worth a look:

• ROS 2 monitoring nodes
• Eclipse Kuksa / VSS-style signal monitoring for structured industrial signals
• SCADE / Simulink Safety / Modelica-based monitoring if you do model-based engineering
• Runtime verification tools such as:
  • Copilot
  • TeSSLa
  • Lustre-based monitors
• Industrial safety systems from vendors like:
  • Pilz
  • SICK
  • Siemens safety PLCs
  • Rockwell safety systems
  • Beckhoff TwinSAFE

Selection criteria

Choose based on:

• Real-time latency requirements
• Safety certification needs
• Integration with robot middleware
• Ability to trigger safe fallback
• Support for temporal/spatial constraints
• Auditability and traceability

My short recommendation

For autonomous industrial robots, the safest practical setup is:

Runtime safety monitor + safety PLC/controller + fallback stop/reduced-speed behavior

If you want, I can also recommend a specific architecture for:

1. ROS 2-based robots,
2. industrial mobile robots/AMRs, or
3. manipulator arms.