Date of Award
Santa Clara : Santa Clara University, 2019
Master of Science (MS)
Adaptive Navigation (AN) has, in the past, been successfully accomplished by using mobile multi-robot systems (MMS) in highly structured formations known as clusters. Such multi-robot adaptive navigation (MAN) allows for real-time reaction to sensor readings and navigation to a goal location not known a priori. This thesis successfully reproduces MAN cluster techniques via swarm control techniques, a less computationally expensive but less formalized control technique for MMS, which achieves robot control through a combination of primitive robot behaviors. While powerful for large numbers of robots, swarm robotics often relies on “emergent” swarm behaviors resulting from robot-level behaviors, rather than top-down specification of swarm behaviors. For adaptive navigation purposes, it was desired to be able to specify swarm-level behavior from a top down perspective rather than experimenting with emergent behaviors. To this end, a simulation environment was developed to allow rapid development and vetting of swarm behaviors while easily interfacing with an existing testbed for validation on hardware. An initial suite of robot primitive and composite behaviors was developed and vetted using this simulator, and the behaviors were validated using the existing testbed in Santa Clara University’s Robotics System Laboratory (RSL). Of particular importance were the adaptive navigation primitives of extrema finding and contour finding and following. These AN primitives were tested over a variety of experimental parameters, yielding design guidelines for top-down specification of swarm robotic adaptive navigation. These design guidelines are presented, and their usefulness is demonstrated for a Contour Finding and Following application using the RSL’s testbed. Finally, possible future work to expand the capability of swarm-based adaptive navigation techniques is discussed.
Metzger, Nathan, "Swarm-Based Techniques for Adaptive Navigation Primitives" (2019). Mechanical Engineering Master's Theses. 37.