Using mixed-initiative human-robot interaction to bound performance in a search task

Mobile robots are increasingly used in dangerous domains, because they can keep humans out of harm's way. Despite their advantages in hazardous environments, their general acceptance in other less dangerous domains has not been apparent and, even in dangerous environments, robots are often viewed as a "last-possible choice." In order to increase the utility and acceptance of robots in hazardous domains researchers at the Idaho National Laboratory have both developed and tested novel mixed-initiative solutions that support the human-robot interactions. In a recent "dirty-bomb" experiment, participants exhibited different search strategies making it difficult to determine any performance benefits. This paper presents a method for categorizing the search patterns and shows that the mixed-initiative solution decreased the time to complete the task and decreased the performance spread between participants independent of prior training and of individual strategies used to accomplish the task.