Why it matters
Without it, mobbing incidents can escalate, birds become stressed or injured, displays are disrupted, and safety licensing checks may flag gaps in risk management.
From design to safety
Design
AI outlines mobbing risk model
Define data sources, thresholds, sensors, and coverage: flight timing beacons, attack history logs, birds present, weather, visibility, daylight, and spotter allocation. Specify guardrails for escalation, repeat alerts, and logging.
- flying grounds fencing
- spotter radios
- spotter vests
Build
Install sensors and data feeds
Deploy 4K daytime and infrared cameras, flight timing beacons every 25 m along the arena, telemetry tags on hawks and corvids, GPS fences, edge processors, and encrypted data pipelines with failover.
- spotter radios
- telemetry tags and receiver
- perches and furniture
Operation
Real-time alerting during flights
Ingest live video and telemetry, compute risk score every 2 seconds, broadcast color-coded alerts to spotters, falconer, and control room, log decisions for post-flight review, and trigger post-flight debrief worksheets.
- flying grounds fencing
- telemetry tags and receiver
- spotter radios
Maintenance
Ongoing calibration and validation
Assess false positives/negatives, recalibrate thresholds quarterly, update attack history with recent events, verify camera alignment, refresh firmware on all sensors, and audit data integrity monthly.
- perches and furniture
- weathering areas
- flying grounds fencing
Safety
Compliance and safeguarding protocols
Maintain Article 10 readiness, incident logging per flight, verify staff training, ensure licenses readable, publish post-event reviews for welfare and regulatory auditing, and archive records per licensing cycle.
- spotter radios
- flying grounds fencing
- perches and furniture
Part by part
- 1
spotter radios
- Watch for
- Receiver failures, reduced signal range, or battery depletion during patrols.
- AI action
- Signal loss detected; switch to backup channel and alert supervisor.
- 2
spotter vests
- Watch for
- Wear drift, torn reflective patches, or improper fit.
- AI action
- Flag for replacement; remind staff to adjust during pre-flight checks.
- 3
flying grounds fencing
- Watch for
- Loose nets, gaps under perimeter, or degraded anchoring points.
- AI action
- Generate maintenance ticket and schedule immediate inspection.
- 4
telemetry tags and receiver
- Watch for
- Tag signal drop, battery wear, or reader lag during flight windows.
- AI action
- Pause flight window; request tag health check and reader calibration.
- 5
perches and furniture
- Watch for
- Loose joints, splinters, or weight-bearing wear.
- AI action
- Flag for repair or replacement and adjust perch load calculations.
- 6
weathering areas
- Watch for
- Wood rot, insect damage, or metal corrosion affecting safety.
- AI action
- Trigger maintenance planning and environmental control checks.
Worked example
Catching mobbing cue before a flight
Illustrative scenario with sample data.
During a scheduled display, AI detects rising mobbing risk from nearby buzzards and corvids as flight begins, prompting a precautionary halt to reassess spacing and broadcast readiness.
Signal in
flightTime=14:25; weather=clear; birdsPresent=4 hawks, 3 crows; history=high corvid activity; spottersActive=true; tagStatuses=OK; arenaPressure=low
System out
alert: mobbing risk high; action: hold flight, issue spotter cue, adjust formation, log incident; follow-up: post-flight review, staff debrief, and record in risk ledger.
Human sign-off: Head falconer approves temporary halt and initiates spotter redeployment and safety briefing.
How to roll it out
Phase 1
Weeks 1 to 2
Baseline deployment of sensors, initial data collection, and training for staff on alert protocols, with a dry-run display rehearsal.
Phase 2
Weeks 3 to 6
Tuning of thresholds, calibration of telemetry, and integration with display schedules, including red-flag scenarios and recovery drills.
Phase 3
Weeks 7 to 12
Operational readiness; full alerting in live flights and documented post-event reviews, plus quarterly licensing doc sync.
Where people stay in charge
- Only a qualified head falconer or avian safety officer may approve flight halts based on AI alerts and situational context.
- All licensing and welfare decisions must be signed by the centre avian vet or licensing inspector.
- Modifications to display timing or species mix require explicit centre leadership sign-off with documented rationale.
Questions owners and crews ask
What triggers mobbing alerts on the flying ground?
AI analyzes flight timing, historical mobbing events, nearby raptor activity, weather, visibility, and spotter input to raise a risk level and prompt actions, logging each trigger with timestamp and location.
Who signs off on flight halts due to AI alerts?
The head falconer or centre avian vet signs off, with licensing inspector notified if regulation requires. A written brief accompanies each halt for audit.
What data feeds are used for alerts?
Live video cues, telemetry signals from birds, spotter reports, and environmental sensors feed the model. Data is time-stamped and cross-checked against history for continuity.
How are false alarms handled in maintenance?
Staff review flags, adjust thresholds, and document justification to prevent unnecessary halts. Post-event analysis improves the model and reduces recurrence by updating training data.