Landing drones on moving targets
By Sally Cole, Senior Editor
Landing a drone on a moving target – and using fuzzy logic to do it – is simply badass.
In a move with intriguing potential for U.S. military applications, a group of researchers at the University of Cincinnati (UC) are using “fuzzy logic” to teach drones to land on moving targets by programming them to make better on-the-fly navigational decisions.
The ultimate goal for aerial drones is to make them autonomous, meaning that the unmanned aerial vehicles will eventually do most or all of their own flying. Drones will need to become autonomous if they’re going to become commercially viable for uses such as making home deliveries to customers, according to Manish Kumar, an associate professor of mechanical engineering for UC’s College of Engineering and Applied Science.
Kumar, working together with Nicklas Stockton, a UC researcher; and Kelly Cohen, a professor of aerospace engineering at UC, considered the difficulty drones have navigating the ever-changing airspace in a study the group presented at the American Institute of Aeronautics and Astronautics SciTech 2017 in January.
First, try to imagine landing a drone on a moving platform or target, such as a U.S. Navy warship pitching around in high seas.
A drone must land “within a designated area with a small margin of error,” Kumar explains. “Landing a drone on a moving platform is a very difficult problem – both scientifically and from an engineering perspective.”
To meet this challenge, UC researchers are applying a concept known as fuzzy logic. It’s the kind of logic people use subconsciously daily. While scientists are concerned with precision and accuracy in all they do, most people get through their day by making inferences and generalities via fuzzy logic. Instead of seeing the world in black and white, fuzzy logic allows for nuances or degrees of truth.
“In linguistic terms, we say large, medium, and small rather than defining exact sets,” Kumar says. “We want to translate this kind of fuzzy reasoning used in humans to control systems.”
Fuzzy logic helps the drone make good navigational decisions amid a sea of statistical noise, he points out. This particular fuzzy logic is called “genetic fuzzy” because the system evolves over time and continuously discards lesser solutions.
The group was able to successfully apply fuzzy logic in a simulation to prove that it’s an ideal system for navigating under dynamic conditions.
Stockton, an engineering master’s student, is doing cool work putting fuzzy logic to the test in experiments to land quadcopters on robots mounted with landing pads at UC’s unmanned aerial vehicle (UAV) Multi-Agent System Research (MASTER) Lab. “This landing project is a real-world problem,” Stockton says. “A delivery vehicle could have a companion drone make deliveries and land itself.”
In case you hadn’t already guessed that the military might be interested in this work, the U.S. Air Force has offered Stockton a federal position to continue his engineering research at Wright-Patterson Air Force Base when he graduates this summer.
Kumar and Cohen are encouraging cutting-edge fuzzy logic/artificial intelligence (AI) work at UC. Nick Ernest, a doctoral graduate and another student of Cohen’s, started an artificial intelligence company, Psibernetix Inc., which demonstrated the power of fuzzy logic last year when a fuzzy-logic-based AI dubbed “ALPHA” bested a human fighter pilot in simulated dogfights.
Retired U.S. Air Force Col. Gene Lee describes ALPHA as “the most aggressive, responsive, dynamic and credible AI I’ve seen to date.”
Compared with other state-of-the-art techniques of adaptive thinking and deep learning, “our approach appears to possess several advantages,” Cohen says. “Genetic fuzzy logic is scalable, adaptable, and very robust.”
UC has become a world leader in fuzzy logic and teaches it at the undergraduate level. “It’s important to introduce students at an early stage to fuzzy approaches because it also provides them with an advantage as they enter the job market,” Cohen notes.
The group’s research was funded by a $500,000 grant from the National Science Foundation.