Semi-Autonomous Vehicle Crashes: An Exploration of Contributing Factors

Abstract

Advancements in semi-autonomous vehicles (SAVs) are amongst the most popular topics in transportation. Proponents of SAVs cite the potential for reducing crashes, roadway congestion, stressful commutes, and increasing independence for persons with disabilities. This enthusiasm is however tempered by crashes involving SAVs. Articles in the popular press cite limitations of the technology, however, crashes rarely have a single cause, and it is important to consider other possible factors. This paper explores the influence of several factors including how SAVs change the responsibility of the driver from operator to supervisor and why drivers are poorly equipped for this supervisory function. We discuss how the opacity of the vehicle’s operation, intentions, and internal states make it difficult for the operators to develop mental models of the systems that enable them to anticipate the automations’ actions. Drivers may instead rely on anthropomorphic reasoning to predict the system’s actions by evaluating how they would respond in a given situation. We explain why anthropomorphisms are problematic and, in some cases, increase crash risk. Finally, we review why drivers operate SAVs in violation of the Operational Design Domain limitations and the contributory role of drivers’ assumptions of the capabilities of SAV technology.

Keywords: Semi-autonomous vehicle, crashes, automation, mental models

DOI: 10.54941/ahfe1002471