An automatic follow drone can save time when you want smooth footage without managing every stick input. The useful part is not just the camera staying on a subject; it is the mix of tracking, obstacle awareness, and flight logic that decides whether the aircraft keeps working when the environment gets messy. In this article I look at how the feature works, where it is genuinely useful, what the UK rules mean in practice, and what I would check before trusting it on a ride, a hike, or a job.
What matters most before you buy or fly one
- The best systems use camera-based subject tracking, not a tracker strapped to the person being followed.
- In the UK, follow-me mode relaxes direct-sight rules only within 50m of the pilot, and the other flight rules still apply.
- For most buyers, obstacle avoidance, reacquisition speed, battery life, and app stability matter more than flashy marketing features.
- Lightweight drones below 250g are easier to use around people, but they still need careful flight planning and privacy awareness.
- If you want the drone for creator work or security tasks, reliability and data handling matter more than camera resolution alone.
What a tracking drone actually does
At its simplest, a tracking drone holds a person, vehicle, or other designated subject in frame while it moves. The software watches the scene, identifies the target, and keeps adjusting altitude, yaw, and position so the subject does not drift out of shot. That is why the feature is useful for solo creators, cyclists, runners, and even some inspection or site-security workflows: it reduces the amount of manual flying you have to do.
There is an important distinction here. Some drones only keep a subject centred and assume the path stays fairly clear. Better systems do more than that: they can plan a route around obstacles, recover the target after a brief loss of view, and behave in a way that feels less like a gimmick and more like a small autonomous assistant. I care about that difference because it is what separates a nice demo from something you can trust in the real world.
That matters most once you move beyond open fields and into places with trees, lamp posts, parked cars, or people crossing through the frame. The more the environment changes, the more the tracking logic matters, which is why the next question is how these systems actually work.
How the subject-following system works
Most consumer tracking features start with the vision system. The drone identifies the subject from the camera feed and keeps measuring its position inside the frame. DJI’s ActiveTrack is a good example of that approach: it tracks a subject visually, and the software can also plan a route and avoid obstacles while it is following. In practical terms, that means the drone is not just “pointing at” the subject; it is actively deciding how to move to keep the shot usable.
Some platforms go further and behave more like autonomous flight systems than simple camera assistants. Skydio’s autonomy stack, for example, is built to keep people or vehicles in frame and reacquire them if they briefly slip out of view. That kind of behaviour is valuable when the subject passes behind a tree, a sign, or a corner, because the drone does not give up at the first visual interruption.
| Layer | What it does | Why it matters |
|---|---|---|
| Vision tracking | Locks onto the subject in the camera feed | No extra tag on the subject, fast to start, easy for casual use |
| Obstacle sensing | Detects objects and adjusts the route | Helps the drone brake or reroute instead of crashing into the scene |
| Reacquisition logic | Searches for the target after a brief loss of view | Makes the feature usable outdoors, not just in ideal conditions |
The key technical point is that the aircraft usually does this without a GPS tracker on the subject. That is convenient, but it also raises the bar for lighting, contrast, and scene clarity, because the drone can only follow what it can actually recognise. Once you understand that, the weak spots become easier to predict.
Where it feels brilliant and where it feels fragile
I think the best way to judge follow mode is by use case, not by spec sheet. It shines when the subject moves at a steady pace and the environment is open enough for the drone to see what is happening. It becomes much less impressive when the subject is hidden, the background is crowded, or the weather is rough.
| Use case | Fit | Main limitation |
|---|---|---|
| Running or cycling on open paths | Strong | Branches, pedestrians, and sudden turns can break tracking |
| Hiking on clear trails | Good | Tree cover and uneven terrain make reacquisition harder |
| Beach, coastal, or open-field footage | Very strong | Wind and glare can still reduce stability |
| Urban streets or crowded parks | Mixed | People, vehicles, and privacy concerns complicate everything |
| Security patrols and inspection work | Strong if professionally managed | Needs permissions, operator discipline, and clear data handling |
The practical lesson is simple: the more the subject can disappear behind something, the more you need a drone that can reacquire quickly instead of freezing or wandering. That is why a follow function is useful, but not magical, and it is also why the UK rules around flying matter so much.
The UK rules that can stop a good shot
In the UK, the CAA’s Drone and Model Aircraft Code is not something to skim and forget. The biggest mistake I see is treating follow mode like a loophole. It is not. It is a narrow exception with clear limits, and the rest of the safety rules still apply.
| Rule | What it means for a follow flight |
|---|---|
| Flyer ID and Operator ID | If the drone weighs 100g or more, you generally need a Flyer ID; if it also has a camera, you need an Operator ID as well. |
| Maximum height | Stay below 120m (400ft) unless a separate permission applies. |
| Distance from people | For many ordinary Open Category flights, keep 50m away from uninvolved people and do not fly over them unless the rules for your sub-category allow it. |
| Follow-me exception | If follow-me mode is active and set to follow within 50m of you, you do not have to keep direct sight at all times during that mode. |
| Privacy | Do not treat recording as a free pass; respect bystanders and avoid unnecessary capture. |
There is also a class-marking angle that matters in practice. Smaller drones below 250g, and certain class-marked models, are easier to use around people than heavier craft, which is one reason so many UK buyers prioritise compact platforms. On the privacy side, the ICO recommends extra transparency measures such as signage and a privacy notice when drones are being used to record people. For me, that is the real professional standard: legal compliance first, then flight performance.
Once those boundaries are clear, the buying decision gets easier, because you can focus on the features that make follow mode actually usable.
What I would prioritise when buying one
I would not start with camera resolution. I would start with the control stack. A drone can shoot sharp footage and still be frustrating if the tracking is flaky, the app lags, or the aircraft gets confused the moment the subject crosses in front of a fence.
| Feature | What I look for | Why I care |
|---|---|---|
| Tracking lock | Fast subject selection and stable lock-on | Setup speed matters when the subject is already moving |
| Obstacle avoidance | At least front, rear, and downward sensing; more is better | Prevents the most common follow-mode failures |
| Battery life | Real-world usefulness, not just the headline figure | I budget for roughly 20-30 minutes of practical flight, not the marketing number |
| Portability | Sub-250g if UK convenience and crowd tolerance matter | Weight changes how often you will actually carry and use it |
| Wind handling | Stable enough for coastal or open-field flying | UK weather punishes weak airframes quickly |
| Recovery controls | Pause, brake, and return-to-home should be obvious | You need a clear escape route when the tracking goes wrong |
| App reliability | Clear menus, responsive tracking, and low lag | The software is part of the product, not an accessory |
My bias is straightforward: I would rather have a drone that is slightly less cinematic but much more predictable. A clean stop button and reliable reacquisition are worth more than a dozen flashy motion modes you will rarely use. That same rule applies when you are setting the drone up for the first flight.
How to get stable results on the first flight
- Start in an open area with no crowds, trees, or narrow gaps.
- Lock the subject at slow speed before you ask the drone to move faster.
- Test pause, brake, and return-to-home before the real shot begins.
- Use a subject with clear contrast against the background, especially in dull light.
- Keep the route simple for the first run; do not begin with corners or sudden elevation changes.
- Watch the battery early and land with margin, not at the last percentage point.
If I am filming a cyclist or runner, I also prefer a route with fewer interruptions and fewer overhead obstructions than most people think they need. The tracking algorithm can be clever, but it still hates ambiguity. If the subject keeps vanishing behind branches, railings, or parked cars, you are forcing the aircraft to do recovery work instead of clean tracking.
The same discipline helps in security or inspection work, where a follow-capable aircraft can save time but only if the operator treats it like an operational tool rather than a party trick. That is where the feature starts to feel genuinely useful instead of merely impressive.
The decision that matters more than the feature list
For solo creators, sports coverage, and some site-security tasks, the point of follow mode is not that the drone is smart. The point is that it reduces friction. It keeps the aircraft doing the boring part while you stay focused on the person, vehicle, or asset that actually matters.
If I were choosing one in the UK, I would favour a lighter, well-behaved drone with dependable tracking, clear recovery controls, and honest obstacle sensing over a heavier model with a louder spec sheet. That approach is less glamorous, but it is the one that keeps paying off after the novelty fades, and it is the difference between a useful tool and an expensive demo.
