Abdollah Homaifar
North Carolina A & T State University, NC·Engineering
Drones increasingly steer themselves using deep learning, AI models that interpret camera and sensor data. These models can be fooled.
- Fields
Engineering
Computer Science
- Topics
unmanned aerial vehicles
drones
adversarial attacks
- Grants
- 1 known grant, through May 2027
What they work on
- Characterize combined trojan and adversarial attacks on deep learning-based UAV navigation, where a model resists ordinary attacks but turns malicious when a hidden trojan trigger is present.
- Develop a deep learning detector that learns from adversarial perturbations and counterfactual attributions, explanations of how outputs shift when inputs change, to detect and mitigate stealthy attacks on drone navigation.
- Investigate secure UAV deployment, including protection of the remote controller and use of secure hardware.
- Evaluate the robustness of navigation models against attacks through simulations and real-world flight experiments on a UAV testbed.
More about the research
Tiny, hard-to-notice changes to what a drone senses can push it off course or hand control to an attacker, which creates real safety risks. This project studies how such attacks work and how to make unmanned aerial vehicle (UAV) navigation systems withstand them.
Why it matters
Drones are used for crop monitoring, search and rescue, and infrastructure inspection, and a hijacked or misguided drone can cause real harm. Methods to detect and resist stealthy attacks on drone navigation could make these systems safer and more reliable for farmers, emergency responders and inspection crews who depend on them.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Jiawei Yuan
University of Massachusetts-Dartmouth, MA·Computer Science
Design and build an integrated, extensible research platform for securely combining AI with unmanned aerial vehicles (drones). The platform is meant to support studies that find weaknesses and test defenses.
Nirnimesh Ghose
University of Nebraska-Lincoln, NE·Computer Science
Develop scalable trust-establishment methods for farm wireless nodes that use chirp spread spectrum (CSS) modulation, a long-range, low-power radio technique used in agricultural IoT.
Joseph Camp
Southern Methodist University, TX·Computer Science
Develop an AI curriculum covering federated learning, explainable AI, homomorphic encryption and neural networks, taught through lectures, exercises and simulations and shared via the NAIRR Pilot.
The school
North Carolina A & T State University: 5 labs on PI Pages, $82M a year on research, 72 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.