They began racing drones. Now these students are using artificial intelligence to build a fleet of them. by Ian Thomsen April 15, 2021 Share Facebook LinkedIn Twitter Northeastern engineering students John Buczek and Noah Ossanna have led a team effort to build a fleet of drones based in artificial intelligence. Photo by Matthew Modoono/Northeastern University A hiker is lost in the wilderness. A large drone, with a dozen skeletal limbs extending out from a frame as large as a household fridge, is launched on a search-and-rescue mission. From a height of 300 feet or more, its artificial intelligence programming dispatches a half-dozen smaller drones from the decks of the mothership. The drones fan out across a large area, locate the missing person, and then regather back on the larger drone. It’s a dream scenario that engineers Noah Ossanna and John Buczek have been building from scratch for the past three years. They’ve been leading a team of fellow Northeastern students to design and build the drone hardware, develop its software, and create a fleet of drones that could perform any number of public services. Their self-made deadline is to fulfill a theoretical mission by next fall on behalf of Aerospace NU, a student club. “To make it truly autonomous, it has to know why and how,” Ossanna says of the swarm technology that will connect the drones as they carry out their group mission. “It has to react in real time.” Buczek, Ossanna, and their teammates at Aerospace NU have continued their three-year mission in spite of the COVID-19 pandemic. Photos by Matthew Modoono/Northeastern University The project goes back to their initial semester at Northeastern, in fall 2017, when Buczek and Ossanna took up drone racing via video goggles connected to a drone camera that creates the perspective of flying. Their racing group expanded to five students, all of whom qualified for the Collegiate Drone Racing nationals in Indiana in spring 2018. “After a while, we were like, ‘OK, this is cool—but what more can we do with drones?’” says Buczek, who will be graduating in December with both a bachelor’s degree and a master’s degree in electrical engineering. “That’s when we started focusing more on research-based projects.” They considered forming their own club before joining Aerospace NU, which is open to all students, regardless of academic background. Buczek, Ossanna, and other team members began teaching themselves how to develop and build drones—they had no experience in the area—and the project took off organically. The more they learned, the higher they reached. The desire to make the drones autonomous introduced them to the Robot Operating System, which enables the fleet of drones to work together. “It’s called ‘swarm carrier,’” says Ossanna, who in 2022 will earn undergraduate and graduate degrees in mechanical engineering and mechatronics (the study of automated manufacturing). “It’s a system to autonomously deploy groups of drones from a larger carrier ship. As sci-fi as that sounds, we have the know-how to do it, and it is actively under construction. “It’s basically this massive aerial vehicle that carries six drones to rapidly deploy them,” Ossanna says. “We’re proposing it as a rapid surveying technique for search and rescue. But the options are limitless. Collaborative autonomy deployed fast is a very compelling technology.” By next fall, the students say, their large drone will carry a fleet of six smaller drones that will be dispatched on a theoretical search-and-rescue mission. Photos by Matthew Modoono/Northeastern University Their project could be applied to surveys of farmlands, bridges and other infrastructure, and wildfire detection that could provide firefighters with real-time information on where a fire is headed. The students designed a system for stacking and deploying the smaller drones on the mothership. The drone at the bottom of its drops down, realizes it’s in free-fall, and starts its motors to fly away—all while the next drone drops down one spot, waiting for its turn. “That is the culmination of two years of research,” Ossanna says of the drop-down design. “I don’t think that method has been employed before. “This isn’t just a bunch of drones doing their own thing,” adds Ossanna. “They each have a computer on board that’s communicating with other drones on the computer network.” Another point of pride is that the proposed mission concludes with the drones completing autonomous landings. “For landing, we have a special stereo camera that senses both depth and infrared, and also a regular RGB [red, green, blue] color camera that we use,” says Buczek, adding that each drone is equipped with an internal mechanism that stabilizes its camera in flight. Ossanna, Buczek, and their clubmates haven’t been overwhelmed by the myriad details they’ve had to learn along the way, says Andrew Gouldstone, a Northeastern professor of mechanical and industrial engineering who serves as adviser to Aerospace NU. “You don’t like to see students bite off more than they can chew, because it can be discouraging,” says Gouldstone. “Noah and John had thought it through. It’s the idea of breaking it down into its deliverables and its challenges—and how are you going to address those? And that allows them to not only complete the project, but it also saves them from the crushing disappointment two weeks into the project where they realize they have to have a flux capacitor.” Each revelation has led to a new set of problems to be explored, understood, and resolved. “We have drones that are, in essence, flying laptops,” Ossanna says. “Each drone can say, ‘OK, I want to go straight—how do I tell my motors to do that?’ But then there’s also basically a laptop on the drone that says, ‘I’m communicating with another computer on another drone that’s telling me to do this, because it sees that.’ “That’s the autonomy that we’re developing, and then we’re also custom-developing the hardware to support it,” Ossanna says. “So it’s very much an interdisciplinary system.” The work has continued in spite of the COVID-19 pandemic—sometimes in a lab on campus, often in their own rooms. Buczek keeps a scaled down version of the carrier ship—known as a dodecacopter, based on its 12 propellers—leaned upright in a corner of his dorm room. “For about a year and a half now, we’ve had a deadline of this coming fall for the total system,” Buczek says. “That’s when Noah and I and many of the other leads are going to convert this into a major capstone project—so we’ll be able to devote even more of our time to it.” The takeaway for Buczek is that exhilaration is a byproduct of setbacks and other frustrations. “For every success, there are however many failures that you have to get past first,” says Buczek. “This project has exposed me to a lot of real-life problems that can be attributed to: There is no bad experience.” “No experiences are bad,” says Ossanna in agreement, “especially when you’re doing it for fun, which is what we’re doing. This project started out as, wow, wouldn’t that be cool … wait, this is actually really complex, let’s see how we can do it. And that’s where we are. We enjoy what we do. And we’re motivated. If you surround yourself with people like that, you end up getting something that’s pretty great.” For media inquiries, please contact firstname.lastname@example.org.