A dock gives you repeatability. The route decides whether that repetition is worth anything. With the Matrice 4D and 4TD in a DJI Dock 3, almost every job reduces to one of three route geometries — waypoint, strip, or area — and the difference between a good and a useless dataset is usually a handful of parameters.
Waypoint routes: inspection of discrete targets
A waypoint route is an ordered list of positions, each with its own altitude, heading, gimbal angle and camera actions. Use it when what you care about is a set of specific things rather than a surface: a substation, a set of chimneys, a row of cell towers, a perimeter with fixed observation points.
What matters:
- Per-waypoint actions. Hover, rotate, pitch the gimbal, shoot wide and zoom, trigger the thermal payload. The value of a waypoint route is that the same frame is captured from the same position every single time, which is what makes change detection possible.
- Heading mode. Locking the aircraft heading to the point of interest gives comparable images; letting it follow the course does not.
- Turn behaviour. Stopping at each point gives sharper stills; curved passes give smoother video and shorter flights.
Strip routes: linear infrastructure
A strip (corridor) route follows a line — a power line, pipeline, railway, road, canal, coastline — with a fixed swath either side. It is a specialised area mission where the geometry is dictated by the asset, not by a polygon you drew.
What matters:
- Corridor width and number of passes. One pass down the centre is rarely enough for anything but visual triage; two offset passes give you both sides of a structure.
- Consistent standoff distance. For thermal and defect detection, distance to the asset drives ground sample distance more than altitude does.
- Segmenting long runs. Corridors usually exceed one battery. Split them into segments that each start and end at a dock-reachable point.
Area routes: mapping and orthomosaics
An area route covers a polygon with parallel passes. This is what you use to produce orthomosaics, digital surface models, stockpile volumes or 3D models.
What matters:
- Overlap. Front and side overlap drive reconstruction quality. For flat, well-textured ground 75/65 is a common starting point; for vegetation, water, uniform surfaces or 3D reconstruction, push both higher and accept longer flights.
- Course angle. Aligning passes with the long axis of the polygon means fewer turns, less battery burnt turning and more consistent lighting across the block.
- Altitude mode. Over sloping terrain, a fixed barometric altitude changes your ground sample distance across the block. Terrain-following keeps it constant — at the cost of needing reliable elevation data.
- Time of day. Fixed-schedule dock missions repeat the same sun angle, which is an advantage for comparison and a problem if that angle is the one with long shadows across your subject.
The parameters that quietly ruin datasets
Four recurring mistakes:
- Overlap set for flat ground and then flown over trees or water.
- Altitude chosen for flight time rather than for the ground sample distance the analysis actually requires.
- Routes edited in place, so last month's imagery and this month's are not comparable. Publish immutable versions and create a new version instead.
- No margin between planned flight time and usable battery, so the mission returns home before the last passes.
Making it repeatable
Once a route produces usable data, freeze it. Give it a version, schedule it, and change it only by publishing a new version. Everything valuable about docked operations — trend lines, change detection, volume differences over time — comes from the fact that the flight is boringly identical each time.