The Continuous Coverage Problem
Coastlines need monitoring 24 hours a day. Smuggling, illegal fishing, migrant vessel distress, environmental spills, search and rescue — the threats do not pause for shift changes or bad weather forecasts. The requirement is simple to state and brutally hard to deliver: unbroken awareness of what is happening along a stretch of coast, every hour, every day.
Current options each fail in their own way. Crewed patrol boats are expensive — fuel, crew salaries, vessel maintenance, insurance, harbour fees — and subject to fatigue. A three-person crew operating 16 hours a day is effective, but the annual cost runs into the hundreds of thousands of euros per vessel. Fixed cameras offer permanence but limited field of view, blind spots between installations, and significant infrastructure cost for coastal mounting, power, and connectivity. Manual drone flights deliver excellent situational awareness for exactly 25 minutes per battery, then nothing. The operator lands, swaps the battery, replans, relaunches. Elapsed time: 5-10 minutes of gap. Multiply that across a full day and the coverage holes are enormous.
The gap between what is needed and what a single drone can deliver is roughly 23 hours and 35 minutes per day. That is not a technology limitation — it is a workflow problem that autonomous systems are built to solve.
A Day in the Life of a Relay Fleet
06:00 — The morning operator arrives at the coastal base station. Opens Overwatch in the browser. Configures a 5 km linear patrol route along the coastline in the planner. Selects 3 drones from the fleet panel, verifies battery states — all green. Clicks launch.
Drone A departs. It climbs to patrol altitude and begins flying the route at cruise speed, camera scanning the water and shoreline below.
06:22 — A hits 20% state of charge. Orchestrate has already anticipated this. At 25% SOC, it dispatched Drone B from the base. B transits to the handoff waypoint — the point on the patrol route where A will run low.
06:24 — B arrives on station and assumes the patrol. A breaks off, returns to base. The operator walks over, swaps A's battery — 60 seconds — and places A back on the pad. A enters the standby queue, fully charged and pre-flighted.
06:46 — B hits the SOC threshold. C dispatches. The cycle repeats.
By 18:00, each drone has completed approximately 15 relay cycles. The patrol route has been continuously covered for 12 hours. The night shift operator takes over, reviews the fleet status, and the patrol continues. It never stopped.
Coverage Gap Math
The critical question is whether the relay creates gaps. The answer depends on transit time.
Transit time from base to handoff point equals distance divided by cruise speed. If the base station is positioned at the coastline midpoint, the furthest handoff point is 2.5 km away. At 5 m/s cruise speed, that is 500 seconds — approximately 8 minutes.
With pre-launch dispatch, the standby drone launches when the active drone hits 25% SOC, not 20%. The ANAFI UKR burns roughly 1% SOC per minute during cruise, so 25% to 20% gives 5 minutes of buffer. The standby drone needs 8 minutes to reach the furthest handoff point. That is a 3-minute deficit on the worst-case leg.
The solution is positioning. For a 5 km linear route, the base at the midpoint means the average transit is 1.25 km — 250 seconds, about 4 minutes. In the average case, the standby arrives at the handoff point before the active drone needs to leave. Overlap is positive. The gap is effectively zero in nominal conditions.
In practice, wind, battery variance, and manual battery swap time introduce small variations. Worst case on the furthest leg: 30-60 seconds of uncovered route. Compare that to 5-10 minutes with a manual swap-and-relaunch approach. The relay reduces the coverage gap by an order of magnitude.
The Reference Fleet
3 ANAFI UKR drones. Why 3: one active, one standby (battery charged, pre-flighted, ready to launch), one in maintenance rotation. The third drone provides redundancy — if one drone develops a hardware fault, the fleet degrades to 2 drones and continues operating at reduced relay frequency rather than stopping entirely. Two-drone operation means slightly longer gaps during maintenance swaps, but the patrol continues.
8 batteries. Why 8: each drone uses approximately 3 batteries per hour in relay mode — roughly 22 minutes of flight time per battery, plus swap and queue time. Over an 8-hour shift, that is 24 battery cycles across the fleet. With a 90-minute charge time and 25-minute flight time, 8 batteries in rotation keeps at least 2 fully charged at all times. The math works out — barely. A ninth battery adds margin for charger delays or a battery that needs to be pulled from rotation due to cell imbalance.
The capital cost of this fleet — 3 drones, 8 batteries, chargers, and a field-deployable base station — is a fraction of a single patrol vessel. The ongoing cost is electricity and battery replacement on cycle count.
What the Operator Actually Does
With exception-only alerting, Orchestrate handles all nominal relay decisions autonomously. The operator's active tasks reduce to: initial setup (5 minutes at shift start), battery swaps (~60 seconds each, every 22-25 minutes, so roughly 3 per hour), and responding to alerts.
The rest is monitoring. One operator can manage the fleet during their shift. The cognitive load is closer to monitoring a security camera system than piloting a drone. No joystick. No video feed to stare at continuously. Swap batteries, acknowledge alerts, maintain situational awareness.
This is a deliberate design choice. The research on human attention degradation during sustained monitoring tasks is clear: performance drops sharply after 20-30 minutes of continuous video watching. By removing the requirement for constant operator attention and replacing it with discrete, event-driven tasks, the system maintains effectiveness across full shifts. The operator stays sharp because they are responding to events, not staring at a feed waiting for something to happen.
Cost Comparison
Patrol boat with a 3-person crew, running 16 hours per day: fuel, crew salaries, vessel maintenance, insurance, harbour fees. Annual operating cost in the hundreds of thousands of euros, depending on vessel class and jurisdiction. A rigid inflatable boat is cheaper than a cutter, but crew costs dominate regardless.
Helicopter patrol hours: EUR 2,000-5,000 per flight hour depending on aircraft type. Annual cost for daily patrols — even short ones — reaches seven figures quickly. Most coastguard units ration helicopter hours precisely because of cost.
Drone fleet: approximately EUR 50,000 capital cost for 3 ANAFI UKR drones, 8 batteries, chargers, and base station equipment. Two operators on 12-hour shifts. Electricity for charging — negligible. Battery replacement every 300-500 cycles, at roughly EUR 150 per battery. Annual operating cost: a fraction of crewed alternatives. The economics are not close.
The drone fleet does not replace the patrol boat or helicopter — it replaces the routine patrol hours that consume the majority of their operating budgets. The crewed assets shift from routine patrol to response, deploying only when the drone fleet detects something that requires intervention. Fleet hours go down. Fuel costs go down. Crew fatigue goes down. Situational awareness goes up.
Limitations
Honest acknowledgment of what this system cannot do today.
Weather. The ANAFI UKR has a maximum wind tolerance of 8 m/s (~29 km/h). That is Beaufort 4. In Beaufort 5+ conditions, the fleet stands down. This is a real operational constraint — coastal environments are windy, and exposed headlands regularly exceed this threshold. The fleet provides coverage when conditions allow, not unconditional 24/7 availability. In many coastal locations, the percentage of flyable hours across a year is 60-80%, depending on season and exposure.
Night operations. The current detection pipeline runs on visible-spectrum imagery, which degrades significantly after dark. The drone can still fly the patrol route at night, but detection capability is limited to well-lit areas or moonlit conditions. Thermal camera integration is on the roadmap and will substantially improve night effectiveness.
Range. The 500 m geofence from the takeoff point limits the patrol radius. For a linear coastal route, this constrains the effective patrol length to approximately 1 km in each direction from base — or longer with waypoint-based geofence extensions, depending on the drone's firmware configuration. For longer coastlines, multiple base stations are needed, each covering its own sector. This is an infrastructure scaling problem, not a software limitation. Orchestrate supports multi-base configurations, but each base requires its own fleet and operator.
These are not theoretical edge cases. They are everyday operational realities. Any team evaluating this system needs to account for them in their coverage planning. A drone fleet is a tool — a very effective one within its envelope — not a replacement for all other assets.