Build the application

## Build
Correlate safety light curtain interruptions and robot joint state data against MoveIt planning scene models. Classify transient zone breaches to formulate targeted physical guard improvements.

## User
Industrial Safety and Ergonomics Specialist

## Workflow
1. Ingest safety PLC trip events and robot Cartesian trajectory records from MoveIt simulation logs.
2. Correlate trip timestamps with robot cell operational phases such as loading, welding, or unloading.
3. Distinguish between nuisance operator boundary interruptions and critical collaborative zone faults.
4. Generate cell safety audit dossiers with proposed safety laser scanner mute zone adjustments.

## Interface
Safety incident triage portal featuring event replay chronologies and zone trip heatmaps.

## Data
Generate synthetic event logs pairing MoveIt joint trajectory points with simulated safety relay status transitions.
Starter examples below are synthetic. They are not downloaded records.

## Source
[MoveIt robotics](https://raw.githubusercontent.com/moveit/moveit2/main/README.md)
Incorporates MoveIt robotics planning scenes and geometric collision bounds to evaluate spatial safety breaches.

## Tools
anthropic-claude, pydantic, pytest.
Hosted Claude provides advanced contextual reasoning over complex multi-variable temporal safety incident logs.
Requires hosted Claude API for structured analysis and safety report synthesis.

## Deliverables
Submit runnable source, pinned dependencies, setup commands, a working interface, example outputs, and a recorded end-to-end demo. Include tests and an action log.

## Acceptance
- Correlates safety trip occurrences with corresponding MoveIt trajectory phase data.
- Categorizes each breach into valid safety classifications according to risk criteria.
- Produces actionable recommendations for protective field zone tuning.
- Flags safety log entries that lack matching motion timestamps.

## Starter Examples
```json
[
  {
    "input": {
      "incident_id": "SAFE-2025-041",
      "cell_id": "WELD-CELL-02",
      "trip_channel": "LIGHT_CURTAIN_ZONE_1",
      "robot_state": {
        "phase": "PART_TRANSFER",
        "velocity_mps": 0.4,
        "distance_to_curtain_m": 0.15
      },
      "operator_tag_present": true
    },
    "expected_output": {
      "incident_id": "SAFE-2025-041",
      "classification": "HUMAN_OPERATOR_EARLY_ENTRY",
      "severity": "MEDIUM",
      "root_cause": "Operator reached into zone before robot reached rest posture",
      "remedial_action": "Extend interlock delay timer by 500ms or adjust entry guard geometry"
    }
  },
  {
    "input": {
      "incident_id": "SAFE-2025-042",
      "cell_id": "WELD-CELL-02",
      "trip_channel": "LIGHT_CURTAIN_ZONE_2",
      "robot_state": null,
      "operator_tag_present": false
    },
    "expected_output": {
      "incident_id": "SAFE-2025-042",
      "classification": "TELEMETRY_UNALIGNED_TRIP",
      "severity": "HIGH",
      "root_cause": "Safety channel tripped without corresponding robot trajectory data",
      "remedial_action": "Audit optical alignment and sensor signal cabling"
    }
  }
]
```

## Safety
Maintains passive analysis and post-incident review; never configures live safety PLCs or safety relays.
Use mock external writes. Require human review before any real deployment.