TECoSA internal research seminar – From Unified Cyber and Adversarial Defense to Efficient Collaborative Perception for Safe Intelligent Autonomous Driving

Registration here: https://www.kth.se/form/69847724a336ea0c04b29934

By Manzoor Hussain, Postdoctoral Fellow at KTH

Abstract

Intelligent autonomous driving systems rely on deep learning–based perception and tightly coupled cyber-physical infrastructures, where low latency, reliability, and efficiency are critical for safety. This talk explores unified and collaborative approaches to enhance the robustness and safety of such systems.

We first present a unified transformer-based framework that simultaneously detects adversarial attacks on perception systems and cyber-attacks on in-vehicle networks, addressing the limitations of deploying separate defense models and reducing computational overhead. Through this unified design, security monitoring is achieved using a single processing pipeline, improving robustness and efficiency in safety-critical environments.

Building on this foundation, we then discuss ongoing work on efficient collaborative perception, where vehicles and infrastructure share sensory information to extend situational awareness beyond local sensing. While motivated by autonomous driving, the challenges addressed including distributed intelligence, low-latency processing, and resource-efficient computation are fundamental to a broader class of edge-enabled AI systems. We conclude by reflecting on how unified security and collaborative perception frameworks can inform the design of future trustworthy and scalable intelligent systems operating at the edge.

Bio: Manzoor Hussain is a Postdoctoral Fellow at KTH Royal Institute of Technology, working within the TECoSA research center. He received his Ph.D. in Computer Science from Chungbuk National University, Republic of Korea, with a thesis titled “A Robust Unified Transformer for Security Resilience in Intelligent Systems.” His research interests include the cyber-security of intelligent systems, AI safety and security, and collaborative perception systems. During his doctoral studies, he worked as a Graduate Research Assistant in the Software Intelligence Engineering Laboratory and completed different National Research Foundation (NRF) funded projects focusing on collaborative cyber-physical systems safety in the autonomous driving domain.