Agent Transparency and Human Performance in the Context of Autonomous Collision Avoidance
摘要
Transparency is a design principle that aims to enhance the understandability and predictability of autonomous systems by depicting their decisions, planned actions, and reasoning to end-users. With transparency, it is anticipated that the availability of this information, directly perceivable on an interface, expedites the information processing for the user and improves situation awareness. However, there is limited research regarding the applicability of transparency in the maritime domain. To address this, a project was initiated to investigate its effects on key human performance variables in the context of autonomous collision avoidance. The project’s main results indicate that improvements in situation awareness can be expected when applying transparency principles, without increasing mental workload. However, future work should investigate transparency’s applicability to time critical applications due to heightened information processing requirements with depicting the system’s analytical processes. Still, the results from this research project showed that transparency is an important prerequisite for supervisory control towards safe and effective autonomous shipping.