HUMAN PERFORMANCE IN INTERPRETING ROBOT-GENERATED 2D AND 3D MAPS

摘要

Robust mapping capabilities are a critical technology for intelligent robotic systems. They can (1) providevaluable information to human and robot teammates without requiring prior knowledge or experience and(2) enable other, higher-level behaviors, such as autonomous navigation and exploration. To maximizeinterpretability, a map must be coherent, accurate, and displayed in an intuitive fashion. However, mapsinherently require a large amount of computational resources. Therefore, it is beneficial to determine theminimum amount of information that must be provided to a user to meet the specific mission requirements.The purpose of this study is to evaluate human performance on visual tasks using 2D and 3D mapsgenerated from laser point cloud data. In a within-subjects study, 20 participants were tasked with locatingand identifying objects, doorways, and windows in a two-story building. The characterizations made hereincould ultimately influence how map data from robotic assets are shared and displayed.