PhD students play major role at Purdue-led AI aviation center

First-year doctoral candidate Jayanth Shreekumar knew of Purdue University’s global reputation because of its strengths in engineering, science and technology.

But a connection to advance his academic path through Purdue happened in an unlikely place: a music concert on campus featuring Indian classical singer and composer Bharat Sundar. There, Shreekumar met Purdue computer science professor Aniket Bera, who introduced him to Sabine Brunswicker, a Purdue professor who leads the new research center, called AIDA³, which is focused on artificial intelligence for autonomous aviation and advanced air mobility.

“I applied to Purdue for two reasons — it is one of the best engineering schools in the world, and my twin brother Ananth is also pursuing his PhD at Purdue in computer science,” says Shreekumar, who started his PhD in electrical and computer engineering this fall. “I was introduced to Professor Brunswicker of AIDA³ by Professor Bera, whom I met at the Sundar music concert. I had a follow-up meeting with Professor Bera regarding my interests, and he referred me to just the right place.”

That place is an exciting role assisting the multi-institutional research team connected with the AIDA³, officially named Center on AI for Digital, Autonomous and Augmented Aviation, which is the first major output of Purdue’s Institute for Physical Artificial Intelligence (IPAI) within Purdue Computes and also is part of Institutes and Centers at Discovery Park District.

With an interest in AI, machine learning and computer vision, Shreekumar is exploring how these tools can be applied to the physical world for enhancing the safety and capabilities of autonomous vehicles. To advance those efforts, Shreekumar is part of an interdisciplinary team of researchers at Purdue and other partnered institutional researchers. Their focus: Develop the necessary infrastructure to establish the nation’s first smart air corridor.

“I am working closely with researchers from different colleges and the technical experts from our industry partnership and enjoying the interdisciplinary collaboration with the team led by Professor Brunswicker,” says Shreekumar, who received a bachelor’s degree from PES University in India and his master’s degree in ECE at UCLA.

“Our goal is to make not only practical impact but also scientific contributions. For example, we plan to develop and innovate new computer vision techniques and algorithms for the AIDA³ vision. I also hope to work on very novel problems in computer vision that will be integrated into the Windracers ULTRA.”

The Windracers ULTRA, named Earhart, is a fixed-wing unmanned aerial vehicle (UAV) manufactured by AIDA³ founding partner Windracers and has a range of 600 miles. UK-based Windracers has donated two ULTRA aircraft — the second is called Armstrong — to be used by AIDA³ for its “from lab to life” research efforts.

Driven by AIDA³, Purdue researchers from five colleges and their collaborators are developing a new scientific field aerial autonomy, called “AIrTonomy.” The field aims to bring together data-driven approaches with models based on mathematical theory and validate the safety and trustworthiness of autonomous vehicles using unique testing facilities. 

To move from digital experiments to physical experimentation and real-world demonstration of the safety of aerial autonomy, AIDA³’s vision is to build the Smart Crossways of America. This planned 200-mile-long air corridor would enable uncrewed aerial vehicles to safely transport cargo, medicine and people from Purdue’s West Lafayette campus to Indianapolis, and the Naval Surface Warfare Center and Muscatatuck Urban Training Center in southern Indiana. 

ADDITIONAL INFORMATION

• Purdue Computes: AIDA³ Doctoral/Postdoctoral Fellowship Program

• Feb. 20 AIDA³ Demonstration Event

• “AIrTonomy: Forming a New Field of Science” video

The brain trust for the smart air corridor is an innovative testing facility known as the Smart Operations Center (SOC), located at the Purdue University Airport. This living laboratory is equipped with a gridded wall of high-definition screens along with motion-capture cameras and VR/AR technologies to facilitate immersive interactions with “outer” spaces. Researchers also can use wearable devices such as high-performing eye-tracking glasses, mobile electroencephalography devices and heart rate sensors.

Since AIDA³ was launched in April, center researchers have focused on demonstrating how an operator physically located at the SOC can perform a remote flight of an uncrewed ULTRA, using research tools and an AI system created by AIDA³.

Currently, it can take 10 people to operate one UAV; the AIDA³ team looks to equip a single operator to simultaneously coordinate up to 100 UAVs. The Purdue center’s goal is to demonstrate the 1-to-10 ratio and to showcase an ability to operate 10 ULTRAs in real urban airspaces with just one operator by the end of 2025.

Shreekumar, a Windracers fellow, is examining how to use computer vision to advance the capabilities of the ULTRA by using cutting-edge AI technology. He’s also examining the mental workload experienced by an operator through various modalities such as webcam, electroencephalography and eye tracker.

“This will be very useful when we want a single operator to control multiple ULTRAs simultaneously. Making sure the operator is not overwhelmed is of the utmost importance,” Shreekumar says. “The eye is the most important organ that connects us to the outside world, allowing us to interact with it. Replicating this for robots is crucial to attain true humanoid robots.”

If successful, AIDA³’s “breakthroughs can truly transform society at scale,” Brunswicker adds.

Purdue PhD student Chuhao Deng, who completed his bachelor’s and master’s degrees in aeronautical and astronautical engineering at Purdue, also is working as a Windracers fellow on Smart Crossways, researching deep learning applications for human-centered air traffic control.

“The application of my current research can make a great impact on how a human interacts with an operation system in safety critical settings,” Deng says. “For instance, the operation system reacts to a human’s cognitive states — such as workload and stress — and a human’s situational awareness, which helps to ensure the system’s safety and efficiency and, in the end, to achieve the 1:100 vision. From there, my next short-term goal is to finish the problem formulation and algorithm development for the research work above, and then to implement that work in real life.” 

An essential part of the planned Smart Crossways is the 15-mile-wide Purdue Unmanned Proving Ground, or PUP. Geographically organized within a triangle near Purdue’s West Lafayette campus, PUP will bring together five unique indoor and outdoor facilities and infrastructure components that ensure research can iterate rapidly from “lab to life,” Brunswicker says.

Other Purdue graduate students working as fellows on the AIDA³ project include Worawis “Willis” Sribunma of aeronautics and astronautics, Doguhan Yeke from computer science, and Liangqi Yuan of electrical and computer engineering. And AIDA³ leaders are embarking on a recruitment campaign to attract more graduate student fellows for the innovative project.

“AIDA³ offers a very unique multidisciplinary opportunity, so fellows get to learn a lot and can grow a broad vision,” Deng says.

Shreekumar adds: “Our center consists of people ranging from various colleges at Purdue — Polytechnic, Engineering, Agriculture, Science and Liberal Arts, collaborating with the Windracers engineers. Our mindset feels very much like that of a startup company. We all learn from each other and come up with exciting innovations.”