Design the mission, then the machine.

The UAM market was exploding in two directions at once: novel vehicle configurations (Joby-style distributed electric air taxis, delivery multicopters) and novel use cases (package delivery, air-taxi service, inspection, search and rescue). But there was no consistent way to go from a compelling use case to a vehicle configured and sized for it. Validated design tools existed only for conventional rotorcraft; startups were burning time on build-test-iterate loops with no guarantee their concept could ever close.

DELIVER's bet was that you could size a working air taxi without an aeronautics degree — and that making the tools that usable was worth coaching for. That coaching was my brief.

The engineers — NASA and DoD — had world-class analytical instincts and technology-first habits. My job was to add one move: start every analysis from the mission and the humans in it, and treat "usable by almost anyone" as an actual requirement.

The deepest HCD foothold in DELIVER was autonomy. You can't size a vehicle without knowing the weight, volume, and power of its autonomy equipment — and you can't know that without designing the mission first. DELIVER's Application-Based Conceptual Design (ABCD) prototype made this concrete: define the use case, derive the autonomy system across three bins (mission systems, concept of operations, regulatory requirements), then feed it into NDARC's sizing as real equipage.

In working sessions, I coached engineers to run that logic the way designers do — personas for operators and bystanders, scenario walk-throughs before equipment lists, "who hears this vehicle and who trusts it" alongside "does the configuration close." The use cases below are the kind we worked through.

Noise is the single largest limiter on urban rotorcraft operations — and the assumption that small electric drones are "quiet" collapses once they operate close to people, in numbers. DELIVER's acoustics work was human-centered research in the strictest sense: free-flight measurements showed multicopter noise is a complex, constantly shifting sum of props at different speeds, and the team built auralizations for human-subject annoyance testing — playing recorded flyovers to listeners and measuring how much the sound actually bothered them, in decibels and in judgments.

The coaching angle: a noise target is a community-acceptance requirement. Annoyance belongs in the conceptual design loop next to range and payload, because noise complaints will shrink where these vehicles are allowed to fly.

DELIVER wasn't a paper exercise — models were calibrated against isolated-prop rigs, wind tunnels, anechoic chambers, and flight campaigns. The rebuilt GL-10 distributed-electric tilt-wing flew a dedicated research campaign in June 2017:

High-fidelity CFD also proved worth running at the concept stage. An OVERFLOW study of prop mounting on a Phantom-like quad caught configuration effects that no rule of thumb would:

* VS FOUR ISOLATED PROPS — PROP–PROP AND PROP–BODY INTERACTIONS. OVER-MOUNT GAINS +2.5% PROP THRUST FROM PSEUDO GROUND EFFECT, LOSES 7.6% TO BODY DOWNLOAD. RELATIVE BAR LENGTHS FOR UNDER-MOUNT CASES ILLUSTRATIVE.

DELIVER demonstrated that NASA's conceptual design tools extend into the UAM concept space — and mapped exactly where they don't — while pulling noise, autonomy, and electrification into the earliest design conversations. The consulting contribution lives in how those conversations ran: starting from use cases, naming the people affected, and treating community acceptance as a hard constraint.

The engineers kept what earned its place in a trade study and dropped what didn't. So I never pitched human-centered design as a virtue. I pointed to where it tightened their numbers and let the trade studies settle it.