Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore a cutting-edge cybersecurity research presentation examining vulnerabilities in Visual Simultaneous Localization and Mapping (vSLAM) systems used in autonomous drones. Learn about the innovative "Phantom Path Attack" methodology that researchers developed to demonstrate how adversaries can manipulate drone navigation by projecting deceptive video stimuli to mislead ORB SLAM3 systems. Discover how this dynamic attack differs from traditional static adversarial inputs by continuously manipulating motion estimation algorithms, causing autonomous drones to deviate significantly from their intended flight paths. Examine the comprehensive evaluation methodology that combines simulation studies with real camera experiments, revealing alarming localization errors of up to 252 meters and altitude deviations of 70 meters that can lead to catastrophic crashes. Understand the broader security implications for vSLAM-based autonomous systems across self-driving vehicles, robotics, and drone applications, while exploring proposed countermeasures including LiDAR/IMU sensor fusion and dynamic filtering techniques to enhance system resilience against adversarial manipulation.
