Analysis of Disengagements in Semi-Autonomous Vehicles: Drivers' Takeover Performance and Operational Implications

2019Final ReportPDFTech Report19-1569A3551747127Driver performanceDriving simulatorsHighway curvesHuman factorsIntelligent vehiclesLevel 2 driving automationLevel 3 driving automationReaction timeTraffic lanesTraffic safetytransportation safetydriftUnited StatesMineta Transportation InstituteFavaro, FrancescaEurich, SkyRizvi, SyedaMahmood, SumaidNader, NazaninMineta Transportation InstituteUnited States. Department of Transportation. Office of the Assistant Secretary for Policy and International AffairsUS Transportation CollectionThis report analyzes the reactions of human drivers placed in simulated Autonomous Technology disengagement scenarios. The study was executed in a human-in-the-loop setting, within a high-fidelity integrated car simulator capable of handling both manual and autonomous driving. A population of 40 individuals was tested, with metrics for control takeover quantification given by: i) response times (considering inputs of steering, throttle, and braking); ii) vehicle drift from the lane centerline after takeover as well as overall (integral) drift over an S-turn curve compared to a baseline obtained in manual driving; and iii) accuracy metrics to quantify human factors associated with the simulation experiment. Independent variables considered for the study were the age of the driver, the speed at the time of disengagement, and the time at which the disengagement occurred (i.e., how long automation was engaged for). The study shows that changes in the vehicle speed significantly affect all the variables investigated, pointing to the importance of setting up thresholds for maximum operational speed of vehicles driven in autonomous mode when the human driver serves as back-up. The results shows that the establishment of an operational threshold could reduce the maximum drift and lead to better control during takeover, perhaps warranting a lower speed limit than conventional vehicles. With regards to the age variable, neither the response times analysis nor the drift analysis provide support for any claim to limit the age of drivers of semi-autonomous vehicles.90