Farcast
Farcast. Landing cargo pods on a target from a moving aircraft.
Farcast builds a hardware and software system that deploys cargo pods from an aircraft mid-flight and lands them on targeted drop zones. I led the iPad app the flight crew relies on to get the release right.
Problem
Very little room for error.
Releasing a pod at the wrong coordinates or altitude lands it outside the drop zone. The aircraft is flying high and fast, and the right release point keeps moving with wind, heading, airspeed and payload.
Meanwhile the pilots are reading instruments, talking to ATC and flying the aircraft safely, all while trying to get as close to the release point as possible.
Research
Learning the cockpit before designing for it.
I shipped TestFlight builds to pilots and gathered their feedback, and spent time in the flight simulator to understand pilot workload and conditions firsthand.
I also flew weekly airdrops as copilot, using the app in the seat it was built for.
Testing where it matters
Filmed from behind the pilot on a live flight, with the iPad mounted beside the instruments running the app.
Constraints
Built for the air, not the office.
Fully offline
The app has to work end to end in the air, with no connection to fall back on.
Two hands on the controls
Pilots can’t afford long interactions, so every action has to be minimal and glanceable.
A moving target
The release point shifts with wind, heading, airspeed and payload, so the app has to update in real time.
Solution
One app for the whole drop.
An all-in-one pilot app with a 3D interface for visualizing the aircraft’s position. It replaced switching between flight-planning tools like ForeFlight.
One glance, full picture
A live 3D view of the aircraft over real terrain beside the 2D route, the drop zone and the next pod, so the crew always knows where they are and what happens next.
Predicting the drop in real time
A machine-learning model reads position, pitch, yaw, speed and wind to predict the trajectory live. I trained it on data from both the flight simulator and real flights to improve accuracy.
Drag, drop, fly
Drag-and-drop waypoints, airport search and a drop-zone editor. Every leg snaps into the route, and the final approach lights up so the crew knows exactly where the release happens.
Real airspace, in view
One tap lays the published air-route network over the map, so every plan is drawn with the real airspace in view.
Impact
More drops on target, less on the pilot’s plate.
Delivery accuracy, validated with pilots and engineers under FAA constraints.
Replaced switching between flight-planning tools in the cockpit, reducing pilot workload.
I changed the app in the aircraft between drops to test improvements in flight, which sharply sped up iteration.
Lessons
What the cockpit taught me.
UIKit over SwiftUI for the map
I first built the mapping in SwiftUI. It was faster to build with, but early SwiftUI was far less performant, so I rebuilt it in UIKit and the difference was noticeable.
Dark screens overheated the iPad
In direct sunlight with heavy GPU use for 3D and mapping, the iPad overheated and the UI glitched. Switching to a light-colored map fixed it.
SceneKit over Mapbox 3D
Mapbox 3D maps didn’t allow setting altitude directly, so I moved to SceneKit and overlaid 3D geography for the target drop zones.