Maze hypothesis development in assessing robot performance during teleoperation

摘要

National Institute of Standards and Technology (NIST) personnel had the opportunity to assess 14 prospective Urban Search and Rescue (USAR) robots, for the purposes of developing performance standards which currently do not exist. During this exercise, a maze configuration -- hypothesized as potentially valid test methodology -- was assessed. Among the findings, resultant significant differences in completion and decision making times facilitated classifying platforms based on performance. Also revealed was that errors in navigation and encounters with walls correlated with times taken in making decisions... the longer it took to make a decision, the greater the chance this decision was incorrect. Results validated the hypothesis of a maze as beneficial in eliciting data necessary for human controlled robot performance assessment.