Enhancing Localization Trustworthiness in GPS-Challenged Autonomous Systems
Date:
Event Details: Daktronics Engineering Hall (DEH 218), South Dakota State University, Brookings, South Dakota
Time: 5:20 PM - 6:20 PM MST (December 5, 2025)
Abstract:
Autonomous vehicles (AVs) and unmanned aerial vehicles (UAVs) rely extensively on GPS for navigation, leaving them exposed to spoofing and other localization attacks that can misdirect autonomous systems and undermine operational safety. This talk presents a structured progression of research aimed at strengthening localization security, beginning with single-sensor approaches that leverage inertial measurements and visual cues to detect inconsistencies between vehicle motion, reported GPS coordinates, and environmental context. Building on the insights gained from single-sensor modalities, the research advances toward multi-modal localization security, integrating information from LiDAR, cameras, IMUs, and geospatial map data to more reliably detect compromised positioning signals, particularly in GPS-challenged or adversarial environments. The talk also highlights emerging attack surfaces in both AV and UAV ecosystems, including vulnerabilities in broadcast identification, cooperative perception, and multi-modal AI models. Collectively, this work demonstrates how fused sensing and AI-driven analysis can significantly enhance the resilience and trustworthiness of next-generation autonomous systems.
Co-sponsored by: EECS, South Dakota State University