MIRA: A Modular Open-Source Micro-UAV for Indoor Research
This paper presents MIRA, a low-cost, open-source, and modular micro-UAV featuring a 3D-printed airframe and containerized software that enables fully customizable indoor research with low-latency communication and vibration characteristics safe for standard control bandwidths.
Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you're trying to build a fleet of tiny, super-smart drones for a science lab. Usually, you're stuck between two bad options: buying expensive store-bought drones that are like locked black boxes (you can't peek inside or change the guts), or building your own from scratch, which takes months of engineering headaches before you even get it to fly.
Enter MIRA (Modular Indoor Research Architecture). Think of MIRA not as a finished product, but as a "Lego set for drone scientists." It's a low-cost, open-source drone designed specifically for indoor research where GPS doesn't work. The big idea? It's built to be modular, repairable, and transparent.
The "Lego" Airframe
The drone's body is 3D printed using PLA (a common plastic). Instead of being one solid piece, it's built like a set of detachable arms. If the drone crashes and snaps an arm, you don't need to throw the whole thing away or solder new wires. You just unscrew the broken arm and snap on a fresh one. It's like changing a tire on a car, but for a drone. The plastic is printed with a special honeycomb pattern inside to keep it light but strong, and to stop the motor vibrations from shaking the brain of the drone too much.
The "Universal Translator" Brain
Most drones have a brain (the autopilot) and a computer (the companion) that speak different languages or are glued together in a way that makes swapping parts impossible. MIRA uses a clever "translator" system. It connects the brain and the computer using a standard digital bridge called Micro XRCE-DDS.
Imagine the autopilot is a pilot who only speaks "Flight," and the computer is a navigator who only speaks "Math." Usually, they need a custom-built walkie-talkie to talk. MIRA uses a universal translator that lets them swap out the pilot or the navigator without ever having to rewrite the instruction manual. This means a lab can use whatever flight controller or computer they already have in their inventory, saving money and time.
The "White-Box" Promise
The authors call this a "white-box" architecture. If a black box is a mystery you can't open, a white box is a clear glass case where you can see every single wire and circuit. This transparency is the main goal: to let researchers focus on their experiments (like testing new AI algorithms) rather than fighting with the drone's hardware.
What the Tests Showed (and Didn't Show)
The team took MIRA for a spin in a room filled with special cameras (an optical motion-capture system) to track its every move. Here is what they found:
- Speed of Thought: The connection between the external cameras and the drone's brain was incredibly fast. The median delay was just 0.02 ms (that's faster than a blink of an eye!). However, they noted that the external camera system itself had some lag, with a median delay of 554 ms and peaks up to 4786 ms. This means the drone was ready instantly, but the "eyes" watching it were a bit slow to send data.
- Shake and Bake: They measured how much the drone vibrated. The motors made the drone shake at a frequency between 90 Hz and 110 Hz. Crucially, the drone's control system operates at frequencies below 20 Hz. Because the vibration is so far away from the control frequency, the drone didn't get confused by its own shaking. The vibrations stayed safely below the safety limits of the autopilot.
- Stability: In "Position Control" mode (where a human pilot holds the drone steady), the drone held its ground well. The vibration levels measured were about 3.5 m/s² for the main sensors and 8.0 m/s² for the secondary ones, which are safe numbers.
What MIRA Is NOT
It's important to know what this paper doesn't claim.
- It's not a race car: The authors explicitly state they are not optimizing for extreme agility or speed. It's a research tool, not a racing drone.
- It's not a magic solution for outdoors: This is strictly for indoor, GPS-denied environments.
- It's not fully autonomous yet: The tests described were done with a human pilot controlling the drone. The paper admits they haven't yet run the "closed-loop" tests where the drone flies itself perfectly on its own.
- It's not perfect: The team found that the external camera system introduced a significant delay (the 554 ms mentioned earlier), which was the biggest bottleneck, not the drone itself. They suggest that in the future, they will move the "vision" processing entirely inside the drone to fix this.
The Bottom Line
MIRA is a solid, open-source foundation. It proves that you can build a drone that is easy to fix, easy to replicate, and easy to swap parts on, all while keeping the software clean and fast. The tests showed it flies stably and handles vibrations well, but the authors are careful to say this is just the beginning. They have a clear path forward: fix the camera lag, get the drone flying itself, and then watch the whole lab fleet take off.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.