Date of Award

Fall 2009

Degree Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Graduate Group

Mechanical Engineering & Applied Mechanics

First Advisor

Mark Yim

Abstract

We present and experimentally verify novel methods for increasing the generality of control, autonomy and reliability for modular robotic systems. In particular, we demonstrate configuration recognition, distributed communication fault tolerance, and the organization and control of self-reassembly with the Connector Kinetic roBot (CKBot). The primary contribution of this work is the presentation and experimental verification of these innovative methods that are general and applicable to other modular robotic systems. We describe our CKBot system and compare it to other similar, state-of-the-art modular robotic systems. Our description and comparison highlights various design developments, features, and notable achievements of these systems. We present work on isomorphic configuration recognition with CKBot. Here, we utilize basic principles from graph theory to create and implement an algorithm on CKBot that automatically recognizes modular robot configurations. In particular, we describe how comparing graph spectra of configuration matrices can be used to find a permutation matrix that maps a given configuration to a known one. If a configuration is matched to one in a library of stored gaits, a permutation mapping is applied and the corresponding coordinated control for locomotion is executed. An implementation of the matching algorithm with small configurations of CKBot configurations that can be rearranged during runtime is presented. We also present work on a distributed fault-tolerance algorithm used to control CKBot configurations. Here, we use a triple modular redundancy approach for CKBot units to collectively vote on observations and execute commands in the presence of infrared (IR) communication failures. In our implementation, we broadcast infrared signals to modules which collaboratively vote on a majority course of action. Various gait selections for a seven module caterpillar and sixteen module quadruped with faulty subsets of IR receivers have been verified to demonstrate the algorithm's robustness. Lastly, we present work on the communication hierarchy and control state machine for the Self-reassembly After Explosion (SAE) robot. Here, we discuss the interaction and integration of the various sensory inputs and control outputs implemented for camera-guided self-reassembly with CKBot. This section describes the overall communication system and reassembly sequence planning after a group of CKBot clusters is kicked apart.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS