Examenspresentation av Darshan Gad

Examiner: Karinne Ramirez Amaro

Opponent: Hampus Ahlebrand

Abstract:
In recent years, the use of robots as the designated solution for simplifying human tasks has seen a notable rise. The incorporation of robots is diverse, and has catered to purely industrial applications like production lines as well as to collaborative environments where humans and robots work together.
However, the introduction of robots into human workflows introduces the challenge of potential errors, caused by software malfunctions, changes in operating environments, or the misinterpretation of human
intentions. Recent studies in the augmented reality domain reveal that providing humans with information about the robot’s intentions, significantly aids in understanding the robot’s behavior and plays a crucial role in building trust. The central objective of this thesis is to investigate the advantages of visualizing robot intentions for users and to conduct a comprehensive assessment of this visualization approach. In the initial phase, we developed a system to enable the visualization of a robot’s intentions, providing users with the capability to replay each action in a robot task. Following this, we conducted a user study to evaluate whether visualizing robot intentions contributes to an enhanced understanding of robot behavior and improves users’ ability to identify robot failures. The results of the user study indicate that users were significantly more successful in understanding the robot’s intentions. When failures occurred, users could identify them 97% of the times by visualizing the robot’s intentions. Moreover, we found that the ability to replay the robot’s actions further enhanced users’ ability in identifying failures in robot actions.

Welcome!

Darshan and Karinne