A drone that carries something eventually has to let go of it. A parcel on a doorstep, a life ring to a swimmer, a sensor pod onto a roof, a marker over a search area, a sling load at a construction site. The carrying is the easy part. The releasing is where the engineering lives, because a mechanism that holds reliably and a mechanism that releases reliably are pulling against each other, and the failure modes on both sides are unpleasant.
Three families of release, and how they differ
| Mechanism | Holding behavior | Power draw | Best fit |
|---|---|---|---|
| Servo with over-center latch | Load carried by geometry, not by the servo | Only while moving | General purpose, light to medium loads |
| Solenoid pin | Pin sheared or held by spring force | Continuous while energized, or pulsed with a latching type | Fast, small, simple releases |
| Electro-permanent magnet | Permanent magnetic hold, toggled by a current pulse | Zero at rest, brief pulse to change state | Repeated cycles, clean ferrous interface, sling loads |
| Motorized winch | Load lowered rather than dropped | Continuous during lowering | Deliveries where the aircraft must stay high |
The servo route is the default for a reason: cheap, well understood, and easy to make fail-safe if the latch is designed so the load path bypasses the servo entirely. That last point is the whole trick. A servo asked to hold a load against its gear train will strip, stall, or drift under vibration. A servo that merely rotates a pin out of an over-center latch, with the load reacted through hardened steel, will run for thousands of cycles. Selecting the actuator on holding torque alone is the most common design error, and the sizing logic in choosing an actuator for a mechanism applies directly.
Electro-permanent magnets are the interesting modern option. They hold with no power at all, which means a total electrical failure does not drop the load, and a brief pulse switches them off to release. The cost is that the payload needs a ferrous interface plate, holding force falls off sharply with any air gap or contamination, and the interface must be kept clean of grit that would otherwise become a hinge under the plate.
The fail-safe question comes first
Before choosing hardware, answer one question in writing: if all power to the release is lost mid-flight, should the payload stay attached or fall? There is no universal right answer, and the mechanism follows from the choice.
For anything with mass over a populated area, the answer is stay attached, and the mechanism must therefore hold passively. A spring-loaded latch that requires energy to open, or an electro-permanent magnet, meets that. A solenoid that holds only while energized does not, and is disqualified for that use regardless of how neat it is.
For emergency jettison, the logic inverts: the crew may need the load gone even after a systems failure, which argues for stored energy in the mechanism, a separate power path, or a mechanical backup. That reasoning is the same family of thinking behind parachute recovery systems, where an independent power source and an independent trigger path are standard practice rather than luxuries.
Whichever direction you choose, write the intended behavior into the requirements and test it by literally cutting power in flight over a safe area. Assumptions about this get made casually and discovered expensively.
Retention: vibration is the real load case
The static load is not what breaks release mechanisms. A multirotor imposes continuous vibration across a wide band plus shock loads at landing, and a parcel swinging on a strap can apply several times its static weight during a gust or an abrupt maneuver. Size the retaining parts for a dynamic multiple of the payload weight rather than the weight itself, and apply an honest factor of safety on the load-carrying elements, because this is a component whose failure endangers people below.
Three details prevent most field problems. Preload the latch so there is no free play, since a latch that rattles will wear its own detent into a ramp and eventually release on its own. Choose spring rates deliberately rather than by feel, because the return spring both retains the latch and drives the release, and spring selection and sizing determines whether it does either reliably at 20 degrees Fahrenheit with ice on it. And design the release so it cannot be assembled in a partially latched state that looks correct on a pre-flight check.
Add sensing. A microswitch or hall sensor reporting latched, unlatched, and payload-present costs almost nothing and turns a hopeful mechanism into a verifiable one. Log the state, and refuse to arm if the mechanism reports unlatched with a payload fitted.
Timing and drop accuracy
Anything released from a moving aircraft continues moving. From 100 feet the fall takes roughly two and a half seconds, and at 30 miles per hour the object travels well over a hundred feet horizontally in that time. Add wind, add drag that varies with orientation, and add the latency between the operator pressing a button and the latch actually opening, and you can see why released packages miss.
Fixing it is mostly software and characterization. Measure the mechanism's actual latency from command to release, including radio link delay, and treat it as a known constant. Compute the release point from ground speed, altitude above the target, and the object's measured drag rather than assuming a vertical drop. Then reduce the problem physically where you can, by hovering rather than flying through, or by descending closer before release. Precision delivery is one of the harder parts of designing an aircraft that carries packages, and it is the reason many delivery designs lower the parcel on a tether instead of dropping it.
Not every release is a discrete drop, either. Continuous dispensing has different physics and different control problems, closer to what spray drones solve with pump and nozzle control, where flow rate has to track ground speed rather than a single trigger event. Payloads released into hazardous environments, as in some firefighting drone applications, add heat and contamination to the same mechanism.
The safety case for releasing anything over people
In the United States, dropping objects from a small unmanned aircraft is addressed directly: 14 CFR 107.23 prohibits allowing an object to be dropped in a manner that creates an undue hazard to persons or property. Operations over human beings sit under Part 107 Subpart D, with categories defined largely by the kinetic energy the aircraft can transfer on impact and by the absence of exposed rotating parts that can lacerate skin. An aircraft that drops things is being evaluated on both the aircraft and the released object.
Build the safety case deliberately. Compute the impact energy of the released object at terminal or release velocity and compare it against the thresholds that define the category you are claiming. Define the exclusion area on the ground and how the operator establishes it. Enumerate failure modes, including inadvertent release, failure to release, partial release with the payload dangling, and release of the mechanism itself. Then design the mitigations: a mechanical safety that requires two independent actions to release, a software interlock tied to altitude and geofence state, and a physical pin for ground handling.
Test to a cycle count well beyond expected service life, at maximum rated payload mass, under vibration, at temperature extremes, and with deliberate contamination. Then test again after a hard landing, because that is the condition under which a real release mechanism eventually gets used.
Projects House designs drone payload release systems, from actuator selection and latch geometry through interlocks and the test program behind them. If your aircraft needs to carry something and put it down accurately, describe the payload and the mission through our contact form.