In GW Engineering’s Department of Electrical and Computer Engineering, current and future engineers, like second-year Ph.D. candidate Rongqian Chen and rising high school senior Allison Andreyev, are taking steps alongside faculty who are at the forefront of shaping emerging technologies. Chen and Andreyev are contributing to key projects led by Professor and IEEE Fellow Tian Lan that leverage artificial intelligence (AI) to create adaptive training simulations and safeguard users of immersive technologies.
Dynamic Duo
In her sophomore year at Wootton High School in Rockville, Maryland, Andreyev was searching for a challenge. She enjoyed coding projects and working with robotics and AI, so she began contacting faculty across Washington, D.C. about research opportunities to explore these topics further. After countless cold emails, she heard back from Lan, who offered her a trial run in his research group to gauge her interest and background while she gained exposure.
“Allison was extremely self-motivated from the beginning, quickly learning new things and never afraid of taking on new tasks. She impressed me during the trial period,” Lan said. “While skills can be developed, motivation is the foundation of success.”
Andreyev began as an intern in November 2024, working closely with Chen, who helped get her up to speed on the latest research in the field. Together, they discussed these advancements and explored future research directions. With a background in robotics and hardware and current doctoral research focused on building intelligent systems to enhance industry and everyday life, Chen was a strong collaborator on Lan’s AI-driven projects.
Merging the Physical and Digital Worlds
Virtual reality (VR), augmented reality (AR), and mixed reality (MR) form a spectrum of immersive technologies blending the physical and digital worlds. While everything in VR is computer-generated, AR incorporates the user’s real-world environment by overlaying virtual elements on the physical world, such as text or markers. In an MR experience, physical and virtual objects coexist and interact in real time.
For high-stakes careers like firefighting and healthcare, incorporating AI into VR offers a unique opportunity to enhance the training of future professionals. As part of a collaboration with Pennsylvania State University (PSU), Lan’s group developed an AI-driven VR healthcare agent that automatically translates training plans into interactive clinical simulations, providing medical students with safe, realistic, and repeatable training.
Chen and Andreyev invited attendees of GW’s InnovationFest to experience this system firsthand in April, demonstrating how users enter a predefined training scenario where the AI agent generated relevant virtual objects, patient cases, and interactive elements based on their decisions. Visitors frequently asked Andreyev if this system would replace traditional training, who emphasized that it was designed to supplement current medical education.
In AR and MR, the combination of real and virtual objects creates avenues for hackers to harm the user. A recent paper, on which Chen and Andreyev are listed as first and second authors, was reported to the Defense Advanced Research Projects Agency as a major accomplishment in the group’s ongoing, multi-institutional project with PSU, Northeastern University, Kennesaw State University, the University of Southern California, and industry partner Design Interactive. With a combined budget of $8.5M, they are collaborating to secure these systems for use in critical mission-driven environments.
For Andreyev, writing content for this project was a major achievement that she feels solidified her as a key contributor. Under the overarching goal of developing novel cognitive models that mimic human perception, the paper focuses on protecting AR against cognitive attacks, which occur when hackers inject malicious objects to alter users’ decision-making.
Traditional attack detection methods are not transparent or explainable, so Chen and Andreyev undertook the arduous task of developing a novel neuro-symbolic approach that combines data-driven patterns, logical reasoning, and contextual knowledge.
“We had to first find all the possible attack patterns, define and classify them, then study how to defend against those. To defend against such attacks, we drew inspiration from human cognition to design a novel framework. This model extracts critical information from the scene and leverages temporal mathematical reasoning to produce interpretable outcomes,” Chen stated.
Andreyev used her coding background to write programs for both parts of this process. “We created programs that would allow an AI agent to place an item into a scene or existing environment to see mathematically how an attack can be detected and what the differences are,” she said.
In collaboration with Northeastern, PSU, and Duke University, the team conducted experiments on an extended dataset and demonstrated improvements in accuracy over traditional methods in challenging AR attack scenarios. Chen and Andreyev shared their work recently at the 2026 IEEE/CVF Conference on Computer Vision and Pattern Recognition, the science and engineering community’s leading annual AI event. Presenting here marks a significant milestone for both, as it is Chen’s first top conference paper at GW and Andreyev’s first opportunity to speak with experts in AI and computer vision outside the university.
Advancing Careers
Now, Andreyev not only has a better understanding of her future career aspirations but also has gained significant hands-on research experience–all before graduating high school. Meanwhile, Chen is already advancing his career goal of enhancing human-machine interaction through his work at GW Engineering on the potential of VR and building systems that can integrate robots or AI agents.
Despite their varied experience levels, Chen and Andreyev showcase how individuals at all stages can meaningfully contribute to groundbreaking research at GW Engineering, bridging the gap between high school ambitions and doctoral-level execution.
The views, opinions and/or findings expressed are those of the author and should not be interpreted as representing the official views or policies of the Department of Defense or the U.S. Government.
Approved for Public Release, Distribution Unlimited.