Gate OS

Gate OS is Turqoa's automated gate transaction module. It processes every truck arrival and departure through a five-stage pipeline — Detect, Verify, Evaluate, Decide, Execute — transforming raw camera feeds into structured, auditable gate decisions in under three seconds.

Architecture

Gate OS runs as a set of containerized microservices deployed on edge nodes at the gate. Each lane operates as an independent processing unit with its own camera cluster, perception pipeline, and decision context.

Cameras → Perception → Validation → Decision
 (3-6)     (OCR +       (TOS +       Engine
            Damage)      Customs)       │
                                        ▼
                         Audit  ←  Orchestrate
                         Trail     (Barrier +
                                    TOS sync)

Transaction Flow

Stage 1 — Detect

A vehicle arriving at the gate triggers the detection stage. Turqoa uses a combination of inductive loop signals (if available) and camera-based motion detection to initiate a transaction.

When triggered, the system captures a synchronized burst of images from all cameras assigned to the lane.

Stage 2 — Verify

The perception pipeline processes captured images in parallel:

• License plate OCR extracts the truck plate number
• Container code OCR extracts the ISO 6346 container ID and check digit
• Seal OCR reads seal numbers from high-security bolt seals
• Damage detection classifies visible damage on container surfaces

Each model returns results with confidence scores that flow into the decision stage.

Stage 3 — Evaluate

The validation layer enriches OCR results by cross-referencing external systems:

• Does the container code match a confirmed TOS appointment?
• Is customs clearance granted for this container?
• Does the truck plate match the appointment's carrier record?
• Are there any watchlist flags on the container or vehicle?

Stage 4 — Decide

The decision engine evaluates its rule set against the enriched data. Possible outcomes:

Stage 5 — Execute

Based on the decision, the orchestration layer:

• Actuates the gate barrier (open or hold)
• Updates the TOS with transaction results
• Sends notifications to relevant parties
• Packages evidence for the audit trail

Camera Requirements

Each gate lane requires a minimum of three cameras:

Note: All cameras must be ONVIF Profile S compliant. Turqoa supports H.264 and H.265 encoding. For plate cameras, infrared illumination is strongly recommended for 24/7 operation.

Configuration

Gate OS is configured through a combination of YAML files and the Turqoa CLI:

# ~/.turqoa/sites/my-terminal/gate-os.yaml
gate:
  lanes:
    - id: lane-01
      direction: inbound
      cameras:
        - lane-01-front
        - lane-01-rear
        - lane-01-plate
      trigger:
        type: motion
        sensitivity: 0.7
        cooldown: 10s
      decision:
        mode: live
        auto_approve: true
        confidence_threshold: 0.92
      barrier:
        controller: modbus://192.168.1.200:502
        open_duration: 30s
        timeout: 60s

Performance Targets

Next Steps

• OCR Engine — Configure and tune OCR models
• Damage Detection — Understand and calibrate damage classification
• Decision Flow — Define gate decision rules and policies