What Follow-Me Mode Actually Does
When you activate follow-me on an AI drone, it locks onto a target — you, a vehicle, a moving object — and keeps it centered in frame while maintaining a set distance and altitude. The drone moves on its own to compensate for wherever the subject goes.
It sounds simple. Under the hood, it involves two quite different technical approaches that behave very differently in the field.
Vision-Based Tracking
Most modern AI drones use computer vision as the primary tracking method. You tap the subject on the app screen, and a neural network classifier assigns that region of the image a unique signature — shape, color pattern, size, gait.
Each new video frame, the processor locates the signature again and calculates how far the subject has moved. Flight controllers adjust motor speeds to compensate and re-center the frame. This happens dozens of times per second.
Vision tracking works well when:
- The subject is clearly visible against a contrasting background
- Lighting is consistent (overcast is often ideal — no harsh shadows)
- The subject is not occluded by trees, buildings or other objects for extended periods
- The drone is within the camera’s effective recognition range — typically 20–60 meters depending on the model
The trade-off: if the subject walks behind a wall, the tracker loses the signature. Some systems re-acquire automatically once the subject reappears; others require a manual retap.
GPS-Based Tracking
Older systems and some dedicated sports trackers use GPS instead of vision. The controller or a wearable beacon broadcasts its GPS coordinates, and the drone flies to stay a fixed distance from those coordinates.
GPS tracking never loses the subject behind an obstacle — it always knows where the beacon is. But it cannot frame the shot as precisely. If the beacon is in a pocket and the person turns around, the drone has no way to know which direction the person is facing.
GPS tracking also depends on signal quality. In canyons, forests or urban environments with tall buildings, multipath errors can cause position jumps.
Hybrid Tracking Systems
Premium AI drones now combine both methods. The drone follows GPS coordinates for gross position, while vision processing keeps the subject properly framed within the shot. If vision locks are lost, GPS maintains proximity until the subject comes back into view.
This combination handles most real-world conditions — forests, crowds, sports scenarios — better than either method alone.
What Affects Tracking Performance
Subject size in frame — too small and the classifier has too few pixels to work with reliably. Fly closer or zoom in digitally.
Subject clothing — plain, solid colors track more reliably than complex patterns. Camouflage is about as difficult as it sounds.
Scene complexity — tracking a cyclist on an empty road is straightforward. Tracking the same cyclist through a crowd challenges even the best systems.
Flight speed limits — follow-me mode typically caps the drone’s speed below its maximum. If you outrun the speed cap, the drone falls behind and may lose tracking.
Lighting transitions — moving from sunlight into deep shade in a second can throw off the classifier.
Common Use Cases
Cyclists, mountain bikers, runners and skiers use follow-me most. The drone captures footage without needing a dedicated operator holding a controller. Kayakers and surfers also use waterproof subject beacons combined with GPS tracking.
For professional shoots — weddings, commercials, sports events — follow-me is a useful autonomous option but is often supplemented by a dedicated camera operator for critical moments.
If you need consistent, reliable tracking, check the active track specifications carefully on any drone you are considering. Our aerial photography use case guide covers what to look for, and our AI drone range shows the tracking capability of each model. Questions about which platform suits your sport or activity? Talk to our team.