How Long the Effect of Take-Over Conditions Lasts? A Survival Analysis of Commercial Motor Vehicle Drivers' Reaction Time and Driving Behavior in Highly Automated Vehicles

Abstract

The transition from automated to manual driving, referred as to Take-over conditions (TOR), in highly automated vehicles (SAE Level 4) is a subject of great interest considering advancement of automotive technologies. However, compared to passenger cars, assessing different aspects of Commercial Motor Vehicle (CMV) drivers’ post-take-over behavior has received less attention, although it is anticipated that CMVs will be the first to vastly adopt highly automated technology. This paper aims to address the question of how long the effect of TOR lasts in CMV drivers and what are the effective parameters. To accomplish this, we designed a 40-minute experiment on a driving simulator and compared participants’ responses to TOR with continuous manual driving to first, assess significant changes in driving behavior indices (e.g., acceleration, velocity, and headway) in different time intervals and second, evaluate the survival patterns of unsafe behaviors (e.g., hard brakes, sharp turns, and speeding) over time. Multilevel Mixed-effect Linear Models and Multilevel Mixed-effect Parametric Survival Models are incorporated to assess the duration of TOR’s effects. Results showed that the first 10s of TOR carries the most significant driving behavior changes while the probability of observing unsafe behaviors reduces significantly after 20s. The results indicated that the effect of TOR lasts longer in long-automated operations, old drivers, and drivers with bad driving history, while repeated TORs, showed positive effects on mediating the effect of this transition. The findings of this paper offer valuable insights to automotive companies and transportation planners on the nature of Take-over conditions.

Keywords: Take-over condition, Commercial motor vehicles, Long-automated operation, Repeated Take-overs, Driver factors, Driving simulator