Safety Implications of Variability in Autonomous Driving Assist Alerting

Mary L. Cummings,Ben Bauchwitz

doi:10.1109/tits.2021.3109555

Mary L. Cummings, Ben Bauchwitz

Open Access

https://doi.org/10.1109/tits.2021.3109555

Copy DOI

Abstract

Advanced Driving Assist Systems (ADAS) are on the rise in new cars, including versions that embed artificial intelligence in computer vision systems that leverage deep learning algorithms. Because these systems, at the present time, cannot operate in all operational driving domains, they employ some type of driver monitoring system for assessing driver attention, so that drivers can effectively take control if and when an ADAS system can no longer control the car. To determine the reliability of a driver alerting system when linked to autonomy that leverages deep learning, a set of increasingly complex tests were conducted on three Tesla Model 3 vehicles. Tests were conducted on a highway and a closed test track to test road departure and construction zone detection capabilities. Results revealed significant between- and within-vehicle variation on a number of metrics related to driver monitoring, alerting, and safe operation of the underlying autonomy. In some cases, cars performed better than expected but all cars exhibited both inconsistent and unsafe behaviors as well as poor driver alerting. These results highlight that a post-deployment regulatory process is ill-equipped to flag significant issues in vehicles with embedded artificial intelligence.

Highlights

M ORE than 92% of new cars sold in the US include some advanced driving assist feature [1], defined as partial automation or Level II Autonomy in the SAE J3016 standard [2]
This paper describes a series of experiments that investigated variability in L2+ driver monitoring alerting in order to reveal potential AI-enabled computer vision and driver monitoring problems within and across multiple vehicles of the same make and model
Three Tesla Model 3 vehicles displayed significant between- and within-vehicle variation on a number of metrics related to driver monitoring, alerting, and safe operation of the underlying autonomy

Summary

Introduction

M ORE than 92% of new cars sold in the US include some advanced driving assist feature [1], defined as partial automation or Level II Autonomy in the SAE J3016 standard [2] Such vehicles are equipped with advanced driver-assist systems (ADAS) that include features like automatic emergency braking (AEB), lane departure warning, and blind spot warning. Some of these cars can perform automated steering and/or acceleration, which many informally call Level II+ (L2+) vehicles. There have been a number of naturalistic studies that indicate people who use L2+ systems spend more time looking in the car when these systems are engaged, and not on the road, increasing the risk of a possible accident [26], [27]

Objectives

Methods

Results

Discussion

Conclusion