Quantifying human behavior in hybrid automation systems using a stochastic approach

The use of hybrid autonomous systems has provides a breakthrough in the design of complex systems in several domains, as it highlights the strength of humans and automation in these systems. However, limited research has focused on developing hybrid autonomous systems, leading to longstanding automation-induced cognitive and performance issues observed in the realm of Human-Automation Interaction. To rectify this, it is imperative to understand the dynamics in task efficiency and human transition behavior within existing adaptive-automated systems. Additionally, modeling these human-automation dynamics via a stochastic approach is essential as it captures inherent human variability within autonomous systems. Therefore, this research explores individual and contextual factors that influence interaction dynamics and performance variability and assesses human transition behavior between automation levels given these factors. The evaluation of these factors involved participants completing a set of tasks in a simulated flight environment and implementing an adaptive automation under varying automation reliability. Additionally, the study entailed the development of a Markovian model, a stochastic model, to quantify the human transition behavior between automation levels in complex systems. Findings from this research offer the potential to identify key design factors relevant in the design of automated systems, and how these factors could ultimately contribute to human-transition behavior in a hybrid-automated system. Ultimately, the inferences of this study could inform designers on how to effectively determine the appropriate automation level for a system to enhance performance, user engagement, and user satisfaction, with the goal of creating a user-centered hybrid automated system.