Small, fast, and nimble, drone racing was a natural transition for many drone hobbyists. The story goes that in 2011, a group of drone hobbyists from Germany decided to set up informal drone races. Posting these races on websites like YouTube led to other groups wanting to organize their own drone races. It didn’t take long for drone racing to become popular all over the world. In 2015, Nicholas Horbaczewski, the former chief revenue officer of the Tough Mudder endurance races, founded the Drone Racing League (DRL). He partnered with Ryan Gury to develop specific FPV (First Person View) drones for the races, along with customized obstacle courses.

The first DRL drone racing season took place in 2016, featuring 5 races and 16 pilot competitors. The races occurred at arenas throughout the United States and were broadcast to 40 different countries by professional networks like ESPN. The 2016 DRL series was won by Jordan “Jet” Temkin from Fort Collins, CO. Jet became the first official professional drone racing pilot, took home a prize of $100,000, and secured a guaranteed spot on the DRL roster for the 2017 season. Drone racing continues to be as popular as ever, with clubs, venues, and championships around the world. The industry was valued at $184.2 million in 2024 and is expected to grow to $602.3 million by 2031.

However, as a team of researchers from Delft University of Technology’s “Cyber Zoo” in Zurich has pointed out, racing drones can extend beyond the sports arena to become vital research tools. Specifically, the team used racing drones to study how drones can be employed to prepare neural networks for AI-enabled spacecraft. Their research was published in June 2024, Volume 9, Issue 91 of Science Robotics, titled “Optimality Principles in Spacecraft Neural Guidance and Control.”

Lead author Dario Izzo explains that the team was tasked with finding a way to test all the onboard resources a spacecraft would need to operate. This includes everything from navigation and object avoidance to available energy resources, propellant, computing, and time management. They knew they needed an AI program to control all the variables, but finding a way to test such conditions in the real world while on Earth was limited. “That’s when we settled on drone racing as the ideal gym environment to test end-to-end neural architectures on real robotic platforms, to increase confidence in their future use in space,” Dario said.

The small drones were set up to run through obstacle courses in the school’s Cyber Zoo. This environment is designed as a 10×10 meter research space surrounded by netting, where users can safely test a wide range of drone and robotic platforms. Surrounding the space are a series of cameras and sensors that collect detailed data on all tests. Sebastien Oringer, another author of the project, explained that spacecraft are controlled by uploading a detailed plan from a ground station for the craft to follow once in space. The goal of this paper was to upload a guidance plan before the drone takes off that would allow it to make necessary corrections once it was airborne.

In space, a spacecraft can encounter various obstacles such as gravitational variations, noise interference, planetary bodies, or turbulence. At this point, the spacecraft needs to adjust the programmed guidance plan to maintain safe control. Without AI, this can be a costly and time-consuming process that can lead to mission failure. The team from Delft designed an AI program and uploaded it to a small racing drone for testing. “Our alternative end-to-end Guidance & Control Networks, G&C Nets, approach involves all the work taking place on the spacecraft,” Sebastien said. “Instead of sticking a single set course, the spacecraft continuously replans its optimal trajectory, starting from the current position it finds itself at, which proves to be much more efficient.”

Though there are clearly differences between drones and spacecraft, the team noted that using racing drones proved to be an ideal starting point for such research. “There’s a whole academic community of drone racing, and it all comes down to winning races,” says Sebastien. “For our G&CNets approach, the use of drones represents a way to build trust, develop a solid theoretical framework and establish safety bounds, ahead of planning an actual space mission demonstrator.” As drone racing continues to push the boundaries of technology and innovation, it not only fuels excitement in the sport but also paves the way for groundbreaking advancements in fields like AI and space exploration.

Previous Drone News: