讲座主题:Deceived Algorithms: Security Blind Spots in Autonomous Driving
主讲人: Prof. Mori Tatsuya, School of Fundamental Science and Engineering, Waseda University
邀请人: Innovation Center of Intelligent Connected Electric Vehicles
时间: 2025.10.24 10:00- 2025.10.24 11:00
地点: Room 200, Yue-Kong Pao Library's Annex
摘要:
With the rapid evolution of autonomous driving technology, AI-driven perception and decision-making systems have become the central nervous system of intelligent vehicles. Deep learning based algorithms for visual recognition, environmental modeling, and path planning enable vehicles to achieve a high level of automation in complex and dynamic environments. However, this heavy reliance on data and models also introduces unprecedented security risks. Security concerns have extended beyond traditional in-vehicle networks to include the algorithmic and sensor physical layers. In the perception phase, adversarial examples can cause deep neural networks to misclassify critical objects such as traffic signs or pedestrians through minimal pixel perturbations or specific sticker patterns, leading to failures in essential tasks. During the training phase, data poisoning attacks can compromise a model by altering a small portion of training samples, implanting backdoors that trigger abnormal decisions under certain conditions. On the physical level, sensor spoofing attacks can create false obstacles or conceal real ones, directly undermining the authenticity and integrity of environmental perception. This lecture will explore the algorithmic world behind autonomous driving and uncover the invisible security threats that lurk within it: when artificial intelligence takes the driver’s seat, hackers and vulnerabilities hide behind the wheel.
主讲人简介:

Tatsuya Mori is a Professor at Waseda University, Japan. He received his Ph.D. in Information Science from the Graduate School of Science and Engineering, Waseda University. His main research areas include information security and privacy protection, AI and machine learning security, autonomous driving and cyber-physical system security, as well as Web3 and cryptographic technology applications.
He is a member of the Information Processing Society of Japan (IPSJ), the Institute of Electronics, Information and Communication Engineers (IEICE), the Japanese Society for Artificial Intelligence (JSAI), the Institute of Electrical Engineers of Japan (IEEJ), and the IEEE. He has previously served as a visiting researcher at the University of Wisconsin–Madison (USA) and a visiting professor at Politecnico di Milano (Italy), and was Chair of the Security Committee of the IPSJ.
In his academic research, Professor Mori has long been engaged in interdisciplinary studies on the design of security mechanisms and adversarial attacks, focusing on systematic defense frameworks that span from the hardware layer to the human factor level. He has published numerous papers in top international conferences such as the USENIX Security Symposium, NDSS (Network and Distributed System Security Symposium), and EuroS&P, among others. His research contributions cover topics such as adversarial robustness of AI models, sensor spoofing attacks and defenses in autonomous driving systems, and privacy-preserving machine learning.