Date of Award
Santa Clara : Santa Clara University, 2016.
Doctor of Philosophy (PhD)
Christopher A. Kitts
Robotics is a transformative technology that will empower our civilization for a new scale of human endeavors. Massive scale is only possible through the collaboration of individual or groups of robots. Collaboration allows specialization, meaning a multirobot system may accommodate heterogeneous platforms including human partners.
This work develops a unified control architecture for collaborative missions comprised of multiple, multi-robot tasks. Using kinematic equations and Jacobian matrices, the system states are transformed into alternative control spaces which are more useful for the designer or more convenient for the operator. The architecture allows multiple tasks to be combined, composing tightly coordinated missions. Using this approach, the designer is able to compensate for non-ideal behavior in the appropriate space using whatever control scheme they choose. This work presents a general design methodology, including analysis techniques for relevant control metrics like stability, responsiveness, and disturbance rejection, which were missing in prior work.
Multiple tasks may be combined into a collaborative mission. The unified motion control architecture merges the control space components for each task into a concise federated system to facilitate analysis and implementation. The task coordination function defines task commands as functions of mission commands and state values to create explicit closed-loop collaboration. This work presents analysis techniques to understand the effects of cross-coupling tasks. This work analyzes system stability for the particular control architecture and identifies an explicit condition to ensure stable switching when reallocating robots. We are unaware of any other automated control architectures that address large-scale collaborative systems composed of task-oriented multi-robot coalitions where relative spatial control is critical to mission performance.
This architecture and methodology have been validated in experiments and in simulations, repeating earlier work and exploring new scenarios and. It can perform large-scale, complex missions via a rigorous design methodology.
Shepard, John T., "A Framework for Collaborative Multi-task, Multi-robot Missions" (2016). Engineering Ph.D. Theses. 2.