How Drone Battery-Swap Docks Work

How Drone Battery-Swap Docks Work: Precision Landing, Hot-Swap Power and Continuous Missions

On 14 October 2026, DJI launched the Dock 4 — its first automated battery-swap drone dock, and the clearest statement yet of where drone operations are going. The headline numbers: a fully automatic battery swap in 25 seconds, and the paired Matrice 5TD aircraft back in the air less than 45 seconds after landing. Compare that with the Dock 3's 27-minute recharge and the operational picture flips — one aircraft can hold near-continuous coverage of a site, a power line corridor or an emergency scene with nobody on site. The dock pairs with the Matrice 5TD (up to 59 minutes of flight, 21 m/s dash speed), rides a sector-shaped battery compartment with four independent bays — three holding charged packs, one kept empty for the depleted pack — and uses induction power to keep the aircraft alive while the battery is out. The whole station is rated IP56, lands in up to 12 m/s wind, and runs from -40 to +50 C.

The product is expensive (the bundle lists at 167,999 yuan, roughly $25,000), but the engineering question underneath it is the interesting part: how do you land a drone precisely enough to grab its battery, swap the pack without rebooting the aircraft, and keep doing it all day? This guide explains each stage — precision landing, the swap mechanism, hot-swap continuity, the battery rack, and the control state machine — with the honest trade-offs, and finishes with the student-scale version we built: the Automated Drone Dock with Battery Swap, a desktop-scale dock with IR homing beacons and a servo-driven swap sled.

Why endurance is the whole game

A quadcopter's flight time is set by a harsh ratio: energy density against the power cost of staying airborne. A mid-size enterprise drone flies 30–59 minutes per pack, and every minute of downtime is a gap in coverage. For drone-in-a-box operations — perimeter security, industrial inspection, emergency response — the aircraft must be ready the instant a call comes in. Recharging on the pad means the aircraft sits through the charge curve; swapping the pack means the downtime is only the mechanism's cycle time. That is the entire economic argument for battery-swap docks: the dock holds N+1 batteries and a charger, so the aircraft is never the thing waiting.

The Dock 4's 4-bay layout is the pattern to copy: three charged packs, one empty slot. The depleted pack goes into the empty slot, the freshest pack goes into the aircraft, the depleted pack starts charging. The rotation is a simple queue, and it is exactly what the student-scale build implements.

Stage 1: Precision landing — getting within centimetres

A swap mechanism cannot reach into a battery tray unless the aircraft is exactly where it expects. GNSS alone is not enough: consumer GNSS positions wander by metres, and even RTK fixes can drift centimetres. Docks layer three mechanisms:

Coarse approach — GNSS/RTK (the Dock 4 uses the DJI O4 Ground Station for its RTK correction) brings the aircraft to roughly above the pad.
Fine homing — infrared beacons on the pad (typically 940 nm LEDs, often modulated at 38 kHz so the receiver can reject sunlight) give the aircraft a directional signal it can centre on. Many docks also use a downward camera reading a landing marker.
Mechanical capture — a funnel or V-guide rim physically centres the landing gear in the last few centimetres, and limit switches or contact pads confirm the aircraft is seated.

For the student build, skip the RTK entirely: a grid of three modulated 940 nm IR beacons on the pad plus an IR receiver on the drone (or the drone flown manually onto the funnelled pad) gets the job done at desktop scale. The honest target is +/- 5 cm alignment — measured with a ruler during bench tests, not claimed from a datasheet.

Stage 2: The swap mechanism — the sled, the tray and the rack

This is the mechatronics core, and the Dock 4's geometry is instructive. The aircraft battery lives in a quick-release tray; the dock carries a linear sled — a motorised carriage on guide rails — that does the exchange in a fixed sequence:

Lock — the aircraft is seated and held; a clamp or the funnel rim prevents shift.
Eject — the sled's fork pulls the depleted pack out of the tray into the empty rack bay.
Index — the rack (or the sled) moves so a charged pack is in line.
Insert — a rack-and-pinion or lead-screw pusher drives the fresh pack into the aircraft tray until the latch clicks.
Verify — the controller reads the pack voltage through the dock contacts before releasing.

Two mechanical details decide whether this works or jams every third cycle. First, the battery tray needs a slight draft angle — 1–2 degrees of taper plus a chamfered leading edge — so the sled never wedges the pack on eject; this is the single most useful builder note in the whole project. Second, the pusher needs stall detection: a servo (an MG996R-class unit delivers about 9.4 kg-cm at 4.8 V, per its datasheet) or stepper should current-limit or time out rather than grind the gears if the pack mis-seats.

Stage 3: Hot-swap continuity — staying alive with no battery

Pull the battery out of a normal drone and everything dies: flight controller reboots, mission state is lost, re-arming takes a minute. The Dock 4 solves this with induction power — coils in the pad energise the aircraft's receiver during the swap so the avionics never see an interruption. That is the genuinely hard part of the product, and the student version should not pretend to replicate it at full fidelity. The honest options, in order of difficulty:

Supercapacitor ride-through — a supercapacitor bank on the aircraft (or in the dock, switched onto the aircraft's power rail) holds the flight controller alive for the 10–30 seconds the tray is empty. Size it from the controller's idle draw, not the motors' draw — the motors are stopped during the swap.
Dual-rail handoff — dock contact pins power the aircraft's low-voltage rail before the pack releases, then the fresh pack takes over. This needs careful sequencing and diode-OR protection so the rails never fight.
Fast swap, accept the reboot — the simplest honest build: the swap is quick enough that losing the flight controller is acceptable, and the state machine re-verifies everything after insertion.

The guide's rule: pick one, name it, and state what you gave up. A dock that honestly reboots is a better project than one that claims seamless hot-swap it cannot demonstrate.

Stage 4: The battery rack — charging while the aircraft flies

The rack is a charger with slots. Each bay holds a pack, monitors it, and reports it to the controller. The student-scale honest design: one TP4056 charge module per slot (1 A charge current, per its datasheet), a per-slot status LED, and a thermistor or simple temperature check that refuses to charge a hot pack. Two honest limitations must be stated on the page, because they decide safety:

TP4056 is a single-cell charger. If the aircraft flies a 2S pack, charging it through single-cell modules is not balance charging — the design must say exactly how the cells are managed (for example: the rack charges via the pack's own balance-charger interface, or the build uses 1S packs). Never present single-cell charging of a multi-cell pack as finished engineering.
Charge current sets turnaround: an 800 mAh pack at 1 A takes the better part of an hour. Three slots and a rotation schedule are what make continuous operation plausible — the arithmetic belongs in the report.

Stage 5: The control state machine

The dock controller (an ESP32 — dual-core, 240 MHz, per Espressif's datasheet) runs one loop that matters: a state machine. LAND_DETECT (weight/contact switch or beacon lock) → ALIGN (beacon readings centred) → LOCK (clamp engaged) → EJECT (sled out) → INDEX (rack to charged bay) → INSERT (pusher in, latch click) → VERIFY (pack voltage read, in range) → RELEASE (clamp off, aircraft cleared to launch). Every transition has a timeout; every timeout has a safe state (usually: stop, LED red, wait for a human). The OLED display (0.96 inch, 128x64) shows the current state, and the Wi-Fi dashboard mirrors it — the same states the web demo on the project page animates.

A short Arduino-style sketch of the core looks like this:

// Dock state machine (simplified). Timeouts omitted for clarity.
enum State { IDLE, ALIGN, LOCK, EJECT, INDEX, INSERT, VERIFY, RELEASE };
State s = IDLE;

void loop() {
  switch (s) {
    case IDLE:
      if (contactSwitchPressed()) { s = ALIGN; oled("Aircraft seated"); }
      break;
    case ALIGN:
      if (beaconsCentred()) s = LOCK;
      break;
    case LOCK:
      clampEngage(); s = EJECT;
      break;
    case EJECT:
      sledEject(); s = INDEX;
      break;
    case INDEX:
      rackToChargedBay(); s = INSERT;
      break;
    case INSERT:
      pusherInsert(); s = VERIFY;
      break;
    case VERIFY:
      if (packVoltageOK()) s = RELEASE;
      else dockFault("Pack voltage out of range");
      break;
    case RELEASE:
      clampRelease(); s = IDLE;
      break;
  }
}

In the real firmware each state gets a timeout and a fault path; the structure above is the spine. Keep it small enough to trace in a viva.

What to measure on the bench

A dock project lives or dies on measured numbers, and the viva will ask for them. Measure: swap cycle time (stopwatch, ten runs, report the range — the design target for the student build is 45–60 seconds, deliberately not DJI's 25); alignment error (ruler, after beacon homing, target +/- 5 cm); eject/insert force margin (does the sled stall? at what pack offset?); contact resistance or voltage drop across the dock contacts under load; charge time per slot from storage voltage to full. Log all of it in a table in the report. These are your numbers — measured, labelled, defensible.

LiPo safety is not optional

The dock lives next to charging lithium batteries, so the safety section of the report is load-bearing: charge only on the rack's monitored slots, never charge a puffed or hot pack, keep a LiPo-safe bag or metal enclosure around the rack, never leave charging unattended, and store packs at storage voltage when the dock is idle. The full treatment is in our LiPo battery safety guide — read it before ordering a single cell.

Frequently asked questions

Why swap batteries instead of just charging the drone on the pad?
Charging keeps the aircraft grounded through the whole charge curve — 27 minutes on the Dock 3, the better part of an hour for a student-scale pack. Swapping moves the waiting onto the rack: the aircraft is down only for the mechanism's cycle time, about 25 seconds on the Dock 4 and 45–60 seconds (design target) on a student build. The rack holds N+1 batteries so there is always a charged pack ready.

How does the dock land the drone precisely enough?
In layers: GNSS/RTK for the coarse approach, infrared beacons or a downward camera for fine homing, and a mechanical funnel that centres the landing gear in the last few centimetres. The student version skips RTK and uses three modulated 940 nm IR beacons plus a funnelled pad, targeting +/- 5 cm alignment measured on the bench.

How does the aircraft stay on during the swap?
The Dock 4 uses induction power from the pad to bridge the gap. Student options are a supercapacitor ride-through sized for the flight controller's idle draw, a dual-rail handoff with diode-OR protection, or — simplest and fully honest — accepting a reboot and re-verifying everything after insertion.

Can a student really build this?
The desktop-scale version: yes. A funnelled landing pad, an IR beacon grid, a servo-driven eject sled, a 3-slot charging rack with TP4056 modules, and an ESP32 state machine is a well-scoped final-year mechatronics build. What you do not replicate is the IP56 enclosure, the RTK precision, or the induction hot-swap — those are stated as limitations. See the Automated Drone Dock with Battery Swap project for the full parts list and build plan.

What does a build like this cost?
A student-scale dock — 3D-printed structure, MG996R-class servos, ESP32, TP4056 rack, IR beacons, OLED, plus three 2S packs — lands in the low five figures in rupees (approximate; the parts list in the project page itemises it). The commercial reference point is the Dock 4 bundle at about 167,999 yuan, roughly $25,000, before software and deployment.

Is automated drone docking legal to test?
In India, DGCA rules govern drone flights; a student dock is a lab prototype tested indoors or tethered at low altitude with the university's permission — not a BVLOS commercial operation. Keep flights indoors or in a netted area, keep the university informed, and never fly near people or airports. The project page lists the exact testing envelope as a limitation.

Honest limitations

The student-scale dock is a lab prototype: no IP56 sealing, no -40 C operation, no 12 m/s wind rating. Those numbers belong to the Dock 4 and stay there.
Swap time is a design target of 45–60 seconds, measured with a stopwatch across ten runs — not DJI's 25 seconds.
TP4056 modules are single-cell chargers; the report must state exactly how multi-cell packs are managed, or use 1S packs.
Autonomous landing is future scope in the base build: the aircraft is flown onto the pad manually (or tethered), and the dock handles detection, alignment verification and the swap.
Contact-based power transfer wears: dock contacts need periodic inspection, and contact resistance should be re-measured across the project's life.
Nothing here is a certified product or a flight-operations manual. Build it as a mechatronics prototype with the battery-safety and flight-safety discipline that implies.

How Drone Battery-Swap Docks Work: Precision Landing, Hot-Swap Power and Continuous Missions