All Projects
UAS SecurityAdversarial MLAssurance

Security and Trustworthiness of UAS

An autonomous aircraft is a target in a way a remotely piloted one is not. An attacker who can shape what its sensors report, or what its learned policy decides, can redirect it without ever touching the control link. GNSS spoofing, perception attacks, poisoned training data, and compromised command links all arrive at the same outcome. This project builds defenses for those attack surfaces and, just as importantly, the assurance evidence an operator or a regulator needs before trusting an autonomous platform with real authority.

Research Objectives

  • Characterize the attack surface of autonomous aerial systems across sensing, learning, and command links
  • Detect GNSS spoofing and sensor manipulation in flight through cross-sensor consistency
  • Harden learned perception and control policies against adversarial and poisoned inputs
  • Define the assurance evidence that justifies granting an autonomous platform operational authority

Methods & Techniques

  • Adversarial robustness evaluation of perception and control models
  • Multi-sensor consistency checking and anomaly detection for spoofing
  • Authenticated and integrity-protected command, control, and telemetry links
  • Runtime verification and structured assurance cases for autonomy certification

Interested in this research?

Get in touch to discuss collaboration or graduate opportunities.

Contact Dr. Hossain