Publication

Autonomy Artificial Intelligence Robotics (AAIR)
2019

ERROR TYPE, RISK, PERFORMANCE, AND TRUST: INVESTIGATING THE IMPACTS OF FALSE ALARMS AND MISSES ON TRUST AND PERFORMANCE

by Huajing Zhao; Hebert Azevedo-Sa; Connor Esterwood; X. Jessie Yang; Lionel Robert; Dawn Tilbury

Abstract

Semi-autonomous vehicles are intended to give drivers multitasking flexibility and to improve driving safety. Yet, drivers have to trust the vehicle’s autonomy to fully leverage the vehicle’s capability. Prior research on driver’s trust in a vehicle’s autonomy has normally assumed that the autonomy was without error. Unfortunately, this may be at times an unrealistic assumption. To address this shortcoming, we seek to examine the impacts of automation errors on the relationship between drivers’ trust in automation and their performance on a nondriving secondary task. More specifically, we plan to investigate false alarms and misses in both low and high risk conditions. To accomplish this, we plan to utilize a 2 (risk conditions) × 4 (alarm conditions) mixed design. The findings of this study are intended to inform Autonomous Driving Systems (ADS) designers by permitting them to appropriately tune the sensitivity of alert systems by understanding the impacts of error type and varying risk conditions.