Adjustable Autonomy for Intelligent Operation of Mobile Manipulators

In recent years, the military has been integrating robotic systems into tasks previously performed entirely manually by soldiers. Examples include explosive ordnance disposal (EOD), surveillance and reconnaissance, urban search and rescue (USAR), chemical, biological, and nuclear (CBN) operations, and medical assistance and evacuation. A typical application is for a soldier (remote operator) to teleoperate a mobile robotic platform (generally coupled with a simple manipulator) that is operating nearby often within direct line-of-sight. Although this technique removes the operator from the acute dangers surrounding the task, the rudimentary nature of current operator interfaces requires intense concentration and makes even simple tasks tedious. As a result, the personal risk to the operator is increased and the full capability of the mechanism is not realized. We propose to design and implement an adjustably autonomous control architecture for complex robot systems that combines robust autonomy with intuitive operator control. This work brings together decades of experience with autonomous control of the highly dexterous Robonaut and decades of experience in remotely operating robots in the field for USAR applications. The resulting system reduces the cognitive load on operators and increases effective utilization of robots in a wide variety of applications. Our dual-armed mobile robotic testbed (TRACBot) has 26 degrees of freedom, LIDAR, stereo, thermal, and color imaging sensors, and will be used for system evaluation and demonstration in several medically-relevant robotics scenarios.