Investigating the Adversarial Robustness of Deep Sensor Fusion Models

In this presentation, Marwen will present use adversarial robustness of fusion model for self-driving car.

title : Can We Trust AI in Self-Driving Cars? Investigating the Adversarial Robustness of Deep Sensor Fusion Models.

Description:

Artificial Intelligence (AI) is transforming the automotive industry, enabling self-driving cars to perceive their surroundings and make complex driving decisions. But how reliable are these AI systems when exposed to malicious attacks?

In this talk, I will present the key findings of my PhD research in Computer and Software Engineering, focusing on the adversarial robustness of deep sensor fusion models for autonomous driving.

I will explore how modern autonomous driving systems combine camera and LiDAR data using deep learning and Transformer-based architectures, how adversarial attacks can compromise their reliability and safety, and how defense strategies can improve their robustness.

Through practical examples and experiments conducted using the CARLA autonomous driving simulator, I will demonstrate how seemingly small manipulations of sensor data can significantly affect the decisions of self-driving vehicles.

The presentation will also highlight the challenges, research findings, and future opportunities for developing safer and more trustworthy AI-powered autonomous systems.

No advanced background in AI or cybersecurity is required. The talk is designed to be accessible to both technical and nontechnical audiences interested in AI, cybersecurity, and autonomous driving.

The presentation will be held at the grande bibliotheque

475 Boul. de Maisonneuve E, Montréal, QC H2L 5C4
room 1.110
The goal of this "informal" event is to learn about AI and discuss it together with the host.