Scalability of Robotic Controllers: Effects of Progressive Levels of Autonomy on Robotic Reconnaissance Tasks
ARMY RESEARCH LAB ABERDEEN PROVING GROUND MD
Pagination or Media Count:
This investigation of the effects of progressive levels of autonomy on robotic reconnaissance task performance was a cooperative effort between the U.S. Army Research LaboratoryHuman Research and Engineering Directorate ARLHRED and Space and Naval Warfare Systems Center San Diego SSC Pacific. Thirty Soldiers served as participants. After training on the operation of the robotic system, each Soldier completed building reconnaissance exercises using three different levels of robotic automation teleoperation, semi-autonomous, and autonomous operation. The participants primary tasks were to map the floor plan of the building, and to identify and photograph objects of interest situational awareness tasks. In order to increase the cognitive demand, Soldiers responded to requests for information while operating the robot. Dependent variables included objective performance data, data collector observations, and Soldier questionnaires. Reconnaissance times were significantly faster and there were significantly fewer driving errors when the robot was in the autonomous mode than when it was in the other two modes. Performance in the semi-autonomous mode was significantly better than in the teleoperation mode. As the level of autonomy increased, workload, reconnaissance times, and driving errors decreased and mapping accuracy increased. There were no significant differences among the conditions in situational awareness or target identification.
- Military Intelligence