· Nimrod Rosenberg · Embedded UAV test pilot

How I Structure a VTOL Flight-Test Campaign, From First Hover to Handoff

A practitioner's breakdown of a VTOL quadplane flight-test campaign: phases, test cards, stop criteria, who does what, and what the engineering team gets after every sortie.

The problem with "let's go fly it"

Most VTOL programs I join already have an aircraft that has flown. Usually it hovered in a car park, maybe did a short forward-flight hop, and someone filmed it. That flight proves the aircraft can leave the ground. It doesn't prove anything the engineering team needs to know next: where the transition is marginal, what the vibration environment looks like at cruise power, how the aircraft behaves when one input is wrong.

A flight-test campaign turns "let's go fly it" into a sequence of questions, each answered with data, in an order that keeps the aircraft and the people around it safe. This post is how I structure one for a VTOL quadplane. The same logic applies to heavy multirotors and conventional fixed-wing; the phases just change.

Start with the questions, not the flights

Before the first card, I sit with the chief engineer and the subsystem owners and ask one thing: what do you need to know by the end of this campaign that you don't know today?

The answers are rarely "does it fly". They sound like:

  • Can it transition with the current pusher at maximum take-off weight?
  • Is the vibration environment acceptable for the payload at cruise?
  • What is the real endurance with the reserve we want to keep?
  • Can a trained operator fly the mission from the procedures alone?

Each of these becomes a test objective. Each objective gets acceptance criteria that we agree on before flying. If we can't write the criterion down, we don't understand the question yet, and flying won't fix that.

I also write down the limits we will not cross: wind, weight, distance from the pilot, minimum battery at landing, and anything the airworthiness owner has restricted. These go on every card.

Phase 0 — Ground and integration

No propellers turning in anger yet. This phase is boring and it is where a lot of campaigns are won.

  • Configuration audit. Parameter file under version control, firmware version recorded, every servo and motor output checked for direction and range on the bench.
  • Failsafe checks on the ground. RC loss, telemetry loss, battery failsafe, geofence. I trigger each one deliberately and check the log shows what we expect.
  • Ground runs. Motors at increasing throttle with the aircraft tied down, looking at the vibration log and current draw. A bad prop or a soft mount often shows up here.
  • Emergency procedures briefed. Who calls abort, how, and what the pilot does. Everyone on the field knows their role before the first takeoff.

Exit criterion: a clean ground-run log, all failsafes verified, and a signed configuration record.

Phase 1 — Multirotor build-up

The quadplane is a multirotor first. I treat it like one.

  1. Low hover in a stabilised mode, short duration, gentle inputs.
  2. Hover with step inputs in roll, pitch and yaw, one axis per card.
  3. Position-hold modes, then small translations.
  4. Weight steps, if the campaign includes higher take-off weights.

At every step I'm reading the attitude tracking (desired versus actual), the motor outputs (how much margin is left) and the vibration levels. If a motor is close to saturation in hover, that is a finding before it's a crash. If the attitude controller needs work, I tune it here, in the field, on the airframe we actually have. I've written separately about why on-site tuning shortens the loop.

Exit criterion: stable hover and translations across the planned weight range, with margins the controls engineer is comfortable with.

Phase 2 — Fixed-wing build-up

Some teams want to go straight to transition. I prefer to understand the wing first, and on many quadplanes you can do that with the lift motors as a safety net.

  • Climb out in multirotor, transition at a conservative point, fly a straight leg in a stabilised fixed-wing mode, then come back.
  • Trim and airspeed calibration. If the airspeed sensor is wrong, everything downstream is wrong.
  • Gentle turns, then pitch and roll doublets, again one axis per card.
  • Stall approach only once the team has agreed how we'll do it and what the recovery is.

Exit criterion: trimmed cruise, calibrated airspeed, and a clear picture of the speed range we'll use for transition work.

Phase 3 — Transition work

This is where VTOL programs spend the most time and where I want the most structure. Forward transition, back-transition, transitions in wind, transitions at weight. Each is its own card, and each one changes only one variable from the last flown point.

I look at altitude loss and gain, airspeed build-up time, attitude excursions, and how hard the lift motors are working as the wing takes over. I cover the tuning side in detail in quadplane transition tuning.

Exit criterion: repeatable transitions in both directions across the agreed weight and wind range, inside the agreed altitude and attitude limits.

Phase 4 — Mission, endurance and degraded modes

Only now do we fly the aircraft the way a customer will: full missions, realistic payload, the reserve we've agreed, and the failure cases. Link loss during cruise. Return-to-launch from distance. Landing with a low battery reserve. Each of these is flown deliberately, at a point in the envelope we've already cleared, so the only new thing is the failure.

Exit criterion: missions flown to procedure, endurance measured rather than estimated, and every planned degraded mode demonstrated.

Phase 5 — Handoff

Handoff isn't the last day. It is a phase where the customer's operators fly from the procedures I've written while I stand next to them. Every time they hesitate, or the procedure says one thing and the aircraft does another, the procedure gets fixed. When they can run a full day without me, we're done.

What the team gets after every sortie

The engineering team should never wait for a test day to be "over" to know what happened. After every sortie, the same evening, they get a short report:

  • What we flew and which card.
  • What happened, with log timestamps.
  • Anything outside expectation, the channel it shows in, and the cause I suspect.
  • What I recommend for the next card.

It's short on purpose. A long report nobody reads before the next morning is worse than a page that changes what we fly tomorrow. My log analysis workflow is what makes these reports quick to produce.

Where campaigns usually go wrong

  • Changing two things at once. New firmware and a new weight on the same card. When something changes, you don't know why.
  • No stop criteria. If nobody agreed what ends a test point, the pilot is deciding in the air. That's the worst place to decide.
  • Software changes between test days that nobody logged. I insist on recording the parameter file for every flight.
  • The pilot reporting feelings. "It felt a bit loose" is a starting point. The report needs the channel and the timestamp.

The short version

A campaign is a structured argument the aircraft makes, one card at a time. My job is to keep each step small enough to be safe, clear enough to be useful, and documented well enough that the engineers can act on it the next morning. If you're planning one, I've also written up what a good test card contains and how to expand the envelope step by step.

Questions

How long does a VTOL flight-test campaign take?

It depends on the aircraft's maturity, the number of open engineering items and how many field days the team can support. I scope it in phases rather than promising a date, and each phase has an exit criterion.

Can we skip the multirotor build-up if the hover controller is inherited from a previous aircraft?

You can shorten it, not skip it. Mass, inertia, motor mounts and vibration paths are new, so the first hovers still need to prove the inherited gains on this airframe.

Who writes the test cards?

I draft them with the engineer who owns each subsystem. The engineer defines what they need to learn; I define how we fly it safely and what stops the test.

Building a new UAV? Book a 20-minute call.