Building a Drone That Flies Itself, Part 1: The Plan

I’m going to build a drone that flies itself. It will take off, fly a long way to a destination or along a route, drop off a package, and come back. Nobody touches the sticks.

I’m writing this down mostly to commit to it. Right now I have an idea, a few spools of filament, and a deadline. That’s it.

Why

A few things I love are finally lining up. I’ve been 3D printing for a while, I write software for a living, and I like building things just to see if I can. A drone I design, print, and program myself touches all three.

It’s also a little crazy, since I have zero drone experience. The closest I’ve come is flying a DJI a few years ago, and that thing basically flies itself. This time I have to build the thing that does the flying.

The goal

First fully autonomous flight within one year.

The mission is simple to describe and hard to pull off:

1. Take off vertically, with no runway and no hand launch.
2. Fly a long distance to a point, or follow a route.
3. Deliver a package.
4. Fly back and land where it started.

The long-term vision is Starship-style: land, recharge, and go again, fully automated and fully reusable. No human in the loop between missions. That’s a way down the road, but it shapes the decisions I make now.

Why a VTOL

A quadcopter is easy to fly and can take off anywhere, but it burns through its battery fighting gravity the whole time. A fixed wing is far more efficient in the air, but it needs space to take off and land.

A VTOL gives me both: it takes off and lands like a multirotor, then transitions to forward flight and cruises like a plane. For a long-range mission that has to come back home on its own, that tradeoff is worth the extra complexity. It’s also the harder option, which is part of the fun.

The stack (so far)

Nothing is final yet, but this is where I’m leaning:

• Flight controller: Matek, probably. I’m also looking at SpeedyBee.
• Firmware: ArduPilot, which has solid VTOL (QuadPlane) support and a huge community.
• Companion computer: a Raspberry Pi running my own code for mission logic and everything that sits above the autopilot.
• Navigation: GPS.
• Links: ELRS for the radio link and telemetry, plus a SIM with internet so range isn’t limited by line of sight to a ground station.
• Airframe: fully 3D printed. I already bought Bambu PLA Aero for the lightweight parts and PETG where I need strength.

As for software, I want to write as much of it myself as I can: mission planning, routing, the delivery logic, monitoring, and whatever else I can think of. That’s the part where I actually know what I’m doing.

The hard parts

This is the list that keeps me up at night:

• Battery. Range is everything, and every gram and every watt counts.
• Range and connectivity. Staying in contact far from home, and knowing what to do when I can’t.
• Regulation. Flying beyond visual line of sight is heavily regulated. I need to figure out what’s allowed here in Uruguay, and where I can legally test.
• Failsafes. An autonomous aircraft has to handle a lost link, a low battery, or bad GPS on its own, safely.
• Weight. A printed airframe that’s light enough to fly far but strong enough to survive landings and a payload.
• Cost. Mistakes in this hobby tend to end in crashes, and crashes cost money.
• Reusability. Landing precisely enough to recharge and launch again without anyone touching it.

Why write about it

Because a public commitment is harder to quietly abandon. This is the first post in a series where I’ll document the whole thing: the designs, the prints, the code, and very likely the crashes.

One year. Let’s see if it flies.