The reason to install a dock is not to fly once. It is to fly the same route, from the same place, on a cadence, for months — and to notice when something changes. Getting there is mostly an operations design problem, not a flying one.
Pick a cadence you can defend
Start from what the data is for:
- Security and situational awareness — several flights a day, short routes, video-first, results reviewed immediately and kept briefly.
- Construction progress — weekly or twice-weekly area mapping, results compared month over month, kept for the life of the project.
- Asset inspection — monthly or quarterly waypoint routes with thermal and zoom captures, kept for years because the value is the trend.
- Stockpile volumes — aligned to the reporting cycle, typically monthly.
Flying more often than you will actually look at the results is the most common waste in docked operations. Every flight costs battery cycles, storage and review time.
Make the schedule concrete
A schedule needs more than a time of day:
- Local time, not UTC. Sun angle is what you are really scheduling against.
- A window, not an instant. Allow the task to start within a tolerance so a brief weather hold does not skip the day entirely.
- A conflict rule. If a dock has multiple scheduled routes, define what happens when they overlap and what has priority.
- A ready-triggered option. For back-to-back missions, launching when the dock reports itself ready beats guessing at fixed intervals.
Let the preflight gate stop the flight
Recurring operations only stay safe if something is allowed to say no. A preflight check should look at battery charge, aircraft health, available storage, wind and precipitation at the dock, and whether the route and airspace clearance are still valid — and refuse the launch with a specific reason when they are not.
Then decide the policy for a refused launch: skip this occurrence, retry within the window, or hold for manual review. Skipping quietly is the option that gets you in trouble; someone should see that the flight did not happen.
Plan for the failure modes
Three things will happen eventually:
- Weather cancels a run. Decide up front whether a missed weekly flight is caught up the next day or simply skipped.
- The network drops mid-mission. The aircraft flies its route and returns home. You lose live video, not the flight. Make sure someone is notified.
- A route becomes invalid. New crane, new structure, changed airspace. Route versions and a scheduled review of altitudes protect against flying yesterday's plan into today's obstacle.
Close the loop on results
A recurring flight is only worth its cost if the output lands somewhere useful: media routed to the right project, a notification when a run completes or fails, and a comparison against the previous run. If reviewing results depends on someone remembering to log in, the programme decays within a couple of months.
Start small
One dock, one route, one daily schedule, watched for a week. Then add the second route. Then tighten the cadence. Autonomy at scale is a series of boring, successful repetitions — not a big-bang rollout.