Lawrence Lei-Ramirez
All work

ARKSEF 2026 · Independent research · Completed

Where a drone frame takes the load.

— five sensors, one very patient ladder

Five load cells, a 3D-printed quadcopter frame, and a five-foot drop — repeated until the data said something a beginner pilot could use.

Instrumented 3D-printed drone test frame with load cells images/arksef-frame.jpg
2400 × 1350 px · 16:9
Fig. 01 — Instrumented test frame, sensors on each arm and at center
Kruskal–Wallis p0.001
Load cells5
Drop height5 ft
Frames cracked2

The question

New pilots crash. Where does the force go?

If impact loading concentrates in a predictable place, frames can be reinforced there instead of everywhere.

Drone flying has gotten far more accessible, which means a lot of first-time pilots putting airframes into the ground during the learning period. Repair cost is the friction point. Knowing how a crash distributes force across the frame is the first step toward a geometry that survives it — or at least fails somewhere cheap to replace.

My hypothesis was that the center hub — where all four arms converge — would carry the highest peak force on impact. Five sensors and a repeatable drop were the way to find out whether that held.

Method

Instrument it, then drop it.

  • 01 · FRAME

    Printed test articles

    Frames were FDM printed so identical articles could be produced cheaply and destroyed repeatedly — the whole study depends on being able to break things without it mattering.

  • 02 · SENSORS

    Five load cells, fixed positions

    One sensor per arm plus one at the center hub, each on an HX711 amplifier, at pre-determined locations so readings stayed comparable across trials.

  • 03 · CAPTURE

    Feather S2 running CircuitPython

    The microcontroller sampled all five channels through impact and logged peak force per sensor per drop.

  • 04 · TEST

    Five-foot drops, repeated

    The instrumented frame was released from a fixed 5 ft height across repeated trials to keep impact energy consistent.

  • 05 · ANALYSIS

    ANOVA and Kruskal–Wallis

    Both a parametric and a non-parametric test were run, since peak-force distributions across sensors weren't safe to assume normal.

Drop test rig wired to the data capture board images/arksef-rig.jpg
1600 × 1200 px · 4:3
Fig. 02 — Drop-test rig and capture electronics
ARKSEF research poster images/arksef-poster.jpg
1600 × 1200 px · 4:3
Fig. 03 — Competition poster

Results

The hub takes it, not the arms.

Relative peak force by sensor position. Sensor 5 sits at the center hub; S1–S4 sit one per arm.

The center hub recorded higher peak force than any arm position, and the gap held across trials.

The Kruskal–Wallis test returned a p-value of 0.001, so the spread between sensor positions is a real effect rather than measurement noise. The hub sat well above all four arms, and no single arm separated itself from the others — consistent with the arms acting as a load path that carries impact energy inward to the point where they meet.

The practical read: extra material earns its weight at the hub and the arm-to-hub junctions. Reinforcing there buys structural integrity that spreading the same mass evenly across the frame does not.

Peak force concentrated at the center, not the arms — so reinforcing the hub is a better use of material than adding mass across the whole frame.

Limits

Drops were vertical and from a single height, so the results describe flat-impact loading rather than the angled, spinning crashes a real pilot produces. Printed PLA also behaves differently from carbon-fiber plate, which most commercial frames use.

What's next

The obvious follow-up is angled drops and a hub-reinforced geometry tested against the baseline frame on the same rig — same sensors, same height, one variable changed.