# ML-based Fault Injection for Autonomous Vehicles: A Case for Bayesian Fault Injection

## TL;DR

This paper introduces DriveFI, a machine learning-based fault injection tool that efficiently identifies safety-critical faults in autonomous vehicle systems, outperforming traditional random testing methods in speed and effectiveness.

## Contribution

The paper presents a novel ML-driven fault injection engine, DriveFI, capable of rapidly discovering safety-critical faults in AV systems, demonstrating significant improvements over random testing.

## Key findings

- DriveFI found 561 safety-critical faults in less than 4 hours.
- Random injection experiments failed to find safety-critical faults over several weeks.
- DriveFI outperforms traditional testing methods in efficiency and fault detection.

## Abstract

The safety and resilience of fully autonomous vehicles (AVs) are of significant concern, as exemplified by several headline-making accidents. While AV development today involves verification, validation, and testing, end-to-end assessment of AV systems under accidental faults in realistic driving scenarios has been largely unexplored. This paper presents DriveFI, a machine learning-based fault injection engine, which can mine situations and faults that maximally impact AV safety, as demonstrated on two industry-grade AV technology stacks (from NVIDIA and Baidu). For example, DriveFI found 561 safety-critical faults in less than 4 hours. In comparison, random injection experiments executed over several weeks could not find any safety-critical faults