MnRI Spotlight: Derya Aksaray
Dr. Aksaray asks, “How do you achieve an efficient communication between human and robot?” and strives to find the answer.
MINNEAPOLIS / ST. PAUL (01/03/2020)—Derya Aksaray is an Assistant Professor in the University of Minnesota’s Department of Aerospace Engineering and Mechanics (AEM). Starting her post-secondary schooling at the Middle East Technical University in Ankara (Turkey), she obtained her M.S. and Ph.D. in Aerospace Engineering at the Georgia Institute of Technology and took some time for post-doctoral work at Boston University and MIT.
After settling in Minneapolis in 2018, Dr. Derya Aksaray has been focusing on her research at the University of Minnesota—resilient autonomy, safe learning, and aerial robotics. We sat down to discuss this research, the advantages the University of Minnesota offers, and the future of her and her team’s work.
“From the institution perspective, the University [of Minnesota] is really good,” Aksaray stated, “it looks like both in the department [of aerospace engineering and mechanics] and in other departments, people are interested in collaborative research and exploring novel ideas… you never feel isolated.”
“A lot of interdisciplinary work?” we suggested.
She responded promptly, saying “exactly, this is one of the best things I like in the University of Minnesota.”
Within Dr. Aksaray’s own lab, she is an advisor for several grad students in AEM. Despite being faculty with AEM, she is also affiliated with the Department of Computer Science and Engineering and works closely with faculty and other students from a handful of departments.
Aksaray and her team work with drones regularly and tackle worldwide industrial problems in their lab nestled within the UMN East Bank’s Akerman Hall. Down in their lab, her student, Ryan Peterson, showed off their new algorithm in a large chess-like demonstration. The floor sectioned off in a grid like pattern, drones simultaneously launched from different spots on the floor and carried their imaginary cargo to another spot. While zipping around the room, they avoid each other (ensuring collision avoidance), prioritizing drop-off for certain drones and ended by landing at specified locations.
Aksaray and Peterson have been working on automating the drone swarm and employing dynamic obstacle avoidance for concepts like urban air mobility/delivery services and have been addressing problems that come with an abundance of autonomous vehicles in a small region—something industry partners are incredibly interested in. Recently, Honeywell has reached out to Aksaray in particular, and they match research interests to make advancements in safe automated flight and beyond.
For autonomous air traffic to benefit large companies, they must be able to operate on a much larger scale. She claims this as her team’s duty to ensure safety, while allowing the A.I. to operate as efficiently as possible (optimizing time, delivery routes, etc.).
“We design algorithms and we theoretically show that they have safety and performance guarantees. We also implement them with some robotic platforms,” Aksaray articulated. The vision for these algorithms start with drone application in urban areas, is tested in laboratories, and ends with being used broadly at Fortune 500 companies, nonprofits, and by the U.S. government. This is to say that Dr. Aksaray and her team’s research is versatile and its full impact could impact many across the world.
Companies like Amazon and UPS have long been looking for opportunities to utilize automated delivery systems and while recently winning the approval of FAA, the research happening in Aksaray’s lab may ensure that flocks of delivery drones have the right programming to safely and efficiently deliver packages to millions of people.
While Dr. Aksaray looks to the future capabilities that drone swarms can have, she is concerned that humans can use them efficiently and effectively. Speaking later about a future interest in heterogeneous teaming, or pairing humans and artificial intelligence to accomplish a single task, she even touched on working with departments outside of the College of Science and Engineering—taking advantage of psychologists to facilitate the pairing of man and machine. The human-robot relationship is at the heart of efficient and safe automation. The solution may come with the help of specialists in human behavior.
“How do you achieve an efficient communication between human and robot?” she asks. She had spent the last months at MIT focusing on how A.I. communicates with humans. Her partners in the behavioral sciences would then observe how humans interact with A.I., through a series of simple puzzle solving tests and more.
“A.I. needs to anticipate what the human is trying to do and… humans should also anticipate about what the A.I. is needing,” Aksaray explained. Just like a well-oiled basketball team, you need coaches who can call plays and react to adverse conditions to pair with players who can execute the game plan and mechanics. While humans can improvise and provide direction, robots with robust programming can execute the task—neither works the best without the other.
While tackling issues regarding automated drone swarms and exploring heterogeneous teaming, they also look for ways to efficiently and safely charge the fleet—which is yet another of the team’s projects and main concerns. With batteries lasting a finite amount of time, they consider mobile charging stations on trucks that deploy fleets of drones for anything from deliveries to emergency relief surveillance. Picture a vehicular commercial aircraft carrier housing and deploying small commercial drones instead of soldiers in F-22 fighter jets.
Although power is essential to the swarms, every part of Aksaray’s research needs to work well in order for drone swarms to work at all for consumer or commercial use. Looking forward, Aksaray and her team hope to continue their work with Honeywell and engage all of the industrial options that the Twin Cities have to offer—many of which “can contribute to robotics research.” The same could be said about Derya Aksaray and the rest of her team at the University of Minnesota.
Dr. Derya Aksaray’s team includes Ali Tevfik Buyukkocak (PhD), Ryan Peterson (MS), Ahmet Semi Asarkaya (MS), and Sophia Vedvik (BS). You can see their work and more on the Control, Robotics, and Intelligent Systems Laboratory website.
