Research & Concepts¶
Before building anything we compared the common ways drones release a payload. The decision is dominated by our platform: a 3.5-inch CineWhoop (SpeedyBee BEE35 Pro) flying indoors. That means:
- Tiny mass budget. The frame already carries a GPS mast, the MTF-01P, a Raspberry Pi Zero 2, an AI camera and an FPV system on 2004-size motors. Every mechanism gram is a gram off the payload and off the hover margin.
- Prop-guard clearance. The ducts sit close to the body; anything that hangs low or swings sideways can touch a duct or shift the CG.
- Indoors = no GPS. Position holding comes from optical flow, which is good to roughly tens of centimetres, not the centimetre-level precision that a docking-style mechanism would need.
- Simple electronics preferred. The companion Pi has free GPIO pins and a 5 V BEC is available; high-current switching hardware is extra weight and extra failure modes.
Options compared¶
1. Servo trapdoor / pin release ← chosen¶
A micro servo either opens a small hatch under the payload bay or pulls a pin that lets the payload slide off a hook. One moving part, one PWM signal.
| Pros | Cons |
|---|---|
| Lightest active option: a 9 g servo plus a printed bracket | One-shot per flight — reloading is done by hand on the ground |
| Trivial electronics: one signal wire to a GPIO, servo power from the existing 5 V BEC | Open-loop: no feedback that the payload actually left |
| Release point is fixed and repeatable — position accuracy is whatever the drone's hover accuracy is, no extra precision needed | Drop height = flight height; the payload falls freely, so it must tolerate the drop (soft payload or low release altitude) |
| Fails safe: an unpowered servo simply holds its last position, the payload stays attached | Mechanism must be designed so vibration cannot creep the pin open |
2. Servo claw / gripper¶
Two jaws (or a single lever) actively clamp the payload and open to release.
| Pros | Cons |
|---|---|
| Can grab irregular payloads; also allows pickup, not just release | Heavier: bigger servo (holding torque under load) plus jaw structure |
| Positive grip during aggressive maneuvers | Servo draws holding current the whole flight — heat and battery drain |
| Jaws protrude below/beside the frame, exactly where duct clearance is tight | |
| Pickup would need centimetre-precision positioning we don't have indoors |
3. Winch lowering¶
The payload is lowered on a line by a motorized spool and released at the end, so the drone can stay high.
| Pros | Cons |
|---|---|
| Gentle delivery — payload touches down at near-zero speed | By far the heaviest and most complex option (motor, spool, line guide, release at the hook) |
| Drone keeps distance from the ground — no downwash interaction at the target | A swinging tethered mass under a 400 g-class quad couples directly into the attitude controller — a real stability risk indoors |
| Line can snag on the prop guards or on obstacles in the hall | |
| Overkill: our delivery height is 1–2 m indoors anyway |
4. Electromagnet¶
An electromagnet holds a ferromagnetic plate on the payload; cutting the current releases it.
| Pros | Cons |
|---|---|
| No moving parts at all; release is instant | Continuous current draw for the whole flight just to hold — the worst possible trade on a small battery |
| Very simple release logic (one MOSFET) | Power loss (brownout, wiring fault) drops the payload immediately — fails dangerous, the opposite of the servo |
| Payload must carry a steel plate (dead mass) | |
| Holding force vs. coil mass scales badly at this size; jerky flight can shear the payload off |
5. Passive hook¶
No actuator: the payload hangs on an open hook and is released by a flight maneuver (touching down, dragging it off) or stays attached until a human takes it.
| Pros | Cons |
|---|---|
| Zero mass beyond the hook, zero electronics, nothing to fail | No commanded release — the mission cannot decide when to drop, which defeats Task 5 |
| Release-by-maneuver needs precise, aggressive flying near the floor — with optical-flow-only position hold that is exactly where the estimate is weakest | |
| Payload can detach unintentionally in any hard maneuver |
Decision¶
The servo pin release wins on every criterion that matters for this platform:
- Mass and clearance — a 9 g servo and a printed bracket are the smallest footprint of any active option, and nothing protrudes into the duct zone.
- Fail-safe behaviour — unpowered = closed. The electromagnet inverts this, the gripper needs constant current, the hook cannot be commanded.
- Electronics we already have — one GPIO pin on the Pi and the existing 5 V BEC. No new power stage, no extra battery load worth measuring.
- Precision match — a fixed release point asks nothing more of the position estimate than hovering over the pad, which is exactly what the vision + optical-flow stack already does.
The mechanism is deliberately one-shot and open-loop: for a graded indoor demo, one reliable drop per flight beats a reloadable mechanism that adds mass and failure modes. The implementation is described in Servo Mechanism; the housing that turns the bare servo into a payload bay is specified in 3D Printing.