SwarmNxt: Open-source Software-Hardware Platform for Fast and Agile Aerial Swarms
This paper introduces SwarmNxt, an open-source software-hardware platform built on the OmniNxt drone that provides an end-to-end toolkit for autonomous aerial swarms, validated through real-world experiments demonstrating high-speed decentralized collision avoidance and collective obstacle-aware flight.
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 a team of first responders rushing toward a collapsed building or a burning forest. In these moments, time is the most scarce resource, and the environment is often too dangerous for humans to enter. A single robot might be able to fly in and look around, but it can only see what is directly in front of it, and if it crashes, the mission ends. A swarm, however, is a group of small flying robots that work together as a single unit. By spreading out, they can cover a large area quickly, map complex spaces, and search for survivors far faster than any individual machine could. The challenge has always been making these groups fly safely on their own. They need to see where they are, avoid hitting each other, and navigate through clutter without crashing, all while running on the small computers built into their bodies. For years, researchers have struggled to build a system that is both powerful enough to think for itself and simple enough for scientists to assemble and test in large numbers.
A team of researchers has now introduced a complete solution to this problem, an open-source platform called SwarmNxt. This system is not just a single drone, but a full toolkit that takes a researcher from the moment they buy the parts to the moment they watch a group of drones fly in a coordinated pattern. The hardware is built on a specific drone design known as OmniNxt, which is small, lightweight, and equipped with cameras that can see in every direction at once. The researchers provided detailed, step-by-step instructions and video guides so that anyone, even without a background in engineering, can assemble these machines. The cost to build one is roughly the price of a high-end laptop, and the assembly takes about five hours. Once built, the real innovation lies in the software. Instead of setting up each drone one by one, which would be slow and prone to error, the team created a system that can configure and update an entire fleet of drones simultaneously. This means that if a researcher wants to test a new flight algorithm, they can push the update to six drones at the same time in just twenty minutes, rather than spending hours on each individual unit.
The heart of the system is a brain that allows the drones to make decisions in real time. Each drone carries a powerful computer capable of processing images to understand the depth of the world around it. Using a learning-based system, the drone looks at pairs of images from its cameras and calculates how far away objects are, creating a 3D map of its surroundings. This map is then used by a planning system that figures out the safest path forward. Crucially, the drones do not just follow a pre-programmed route; they talk to each other to avoid collisions. If one drone sees another getting too close, it adjusts its path instantly. The researchers tested this system in two different scenarios. In the first, they sent six drones into an empty room to see if they could fly in tight formation without hitting one another. The drones swapped positions and roamed around a large circle, flying for a total of two hours without a single collision, despite occasional delays in their wireless communication. In the second test, they reduced the group to four drones and filled the room with obstacles. These drones had to use their onboard cameras to see the walls and furniture, calculate their distance, and navigate through the clutter. They flew for thirty minutes, successfully avoiding both the obstacles and each other.
The results show that the system works reliably, but the researchers are careful to note its current limits. The drones fly best when they are in a room where their position can be tracked by an external camera system, which acts as a perfect guide for their location. They cannot yet fly this way in a GPS-denied outdoor environment without that external help. Furthermore, the system relies heavily on the drone's graphics processor to calculate depth, which takes up most of the computer's power, leaving little room for other complex tasks at the same time. The depth calculation is also limited in resolution, meaning very small objects far away might be missed. Despite these constraints, the platform represents a significant step forward. It proves that a group of small, affordable robots can be assembled, programmed, and flown together with a high degree of autonomy. By making the hardware and software open to everyone, the researchers hope to remove the barriers that have slowed down progress in this field, allowing scientists to focus on solving the harder problems of making these swarms truly independent and ready for the real world.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.