[Physical Adverse Weather (Rain)] + [Adversarial Spoofing (10-20 Points)] │ ▼ [Traditional ML Models Misled / Blinded] │ (Dr. Capraru's Countermeasures) ▼ [Robust Perceptual Defenses & Anti-Forgetting ML] 1. Unmasking LiDAR Vulnerabilities in Adverse Weather
: He has held visiting positions at prominent institutions, including Korea University, the Hong Kong University of Science and Technology (HKUST), Peking University, and the University of Tokyo.
(UCL), where he earned his Bachelor of Engineering. During his time at UCL, he was recognized as a Laidlaw Scholar richard capraru
Richard Capraru’s research is crucial for the automotive and AI industries, which are under pressure to ensure that self-driving cars can operate safely in all environments, including those with adverse weather and potential cybersecurity threats. By identifying how attackers can leverage weather to mask their efforts, this research helps shape the development of more robust, secure sensory technology.
GhostLite: Data Minimization with Applications to Real-Time LiDAR Attacks (UCL), where he earned his Bachelor of Engineering
is an international artificial intelligence researcher specialized in robust autonomous systems, adversarial perception, and cybersecurity in AI. He is affiliated with the International Research Center for Neurointelligence (IRCN) at the University of Tokyo.
, enabling systems to learn new gestures from a minimal number of examples. Semantic Scholar Safety in Autonomous Systems secure sensory technology.
| Year | Title | Publication/Conference | Key Focus | | :--- | :--- | :--- | :--- | | 2020 | Dop‐NET: a micro‐Doppler radar data challenge | Electronics Letters | Creating a shared database for radar-based classification. | | 2020 | Exploring gesture recognition with low-cost CW radar modules | IEEE International Radar Conference (RADAR) | Developing low-cost radar for human-computer interaction. | | 2024 | Rain-Reaper: Unmasking LiDAR-based Detector Vulnerabilities in Rain | IEEE/RSJ IROS | Novel attack exploiting rain to fool LiDAR systems. | | 2025 | Overcoming Catastrophic Forgetting in Radar and Lidar Object Detection in Rain | IEEE Radar Conference | Using ML techniques to maintain detection performance in rain. | | 2025 | GhostLite: Data Minimization with Applications to Real-Time LiDAR Attacks | IEEE Vehicular Technology Conference | Creating efficient and stealthy "ghost" objects. | | 2026 | Leveraging Adverse Weather for Enhanced LiDAR Spoofing in Autonomous Driving | IEEE Vehicular Technology Magazine | Broader analysis of weather's role in sensor spoofing. |