Towards a Systematic Risk Assessment of Deep Neural Network Limitations in Autonomous Driving Perception

TL;DR

This paper presents a systematic risk assessment workflow for identifying and analyzing risks from deep neural network limitations in autonomous driving perception, combining hazard and threat analysis methods.

Contribution

It introduces a joint workflow integrating ISO standards to systematically evaluate DNN limitations' risks in autonomous driving systems.

Findings

Proposes a combined hazard and threat analysis workflow.

Addresses the lack of systematic risk assessment for DNN limitations.

Highlights potential safety and security risks in autonomous driving perception.

Abstract

Safety and security are essential for the admission and acceptance of automated and autonomous vehicles. Deep neural networks (DNNs) are widely used for perception and further components of the autonomous driving (AD) stack. However, they possess several limitations, including lack of generalization, efficiency, explainability, plausibility, and robustness. These insufficiencies can pose significant risks to autonomous driving systems. However, hazards, threats, and risks associated with DNN limitations in this domain have not been systematically studied so far. In this work, we propose a joint workflow for risk assessment combining the hazard analysis and risk assessment (HARA) following ISO 26262 and threat analysis and risk assessment (TARA) following the ISO/SAE 21434 to identify and analyze risks arising from inherent DNN limitations in AD perception.