A reconfigurable cooperative control system for rapid deployment of multi-robot systems.

摘要

Multi-robot systems have been used in a vast array of fields and are of particular interest in perilous environments. Utilizing multiple smaller and cheaper robots have many advantages compared to a highly specialized single robot. Multi-robot systems are fault-tolerant by nature and provide task completion parallelism for faster mission completion. One of the main issues in multi-robot systems is the lack of a common set of abstractions and middleware. Controlling and programming cooperative multi-robot systems is a highly complicated task that requires a flexible and agile control architecture and programming environment that are able to handle the distributed nature of multi-robot system. This dissertation studies many different aspects of multi-robot systems. The major characteristics, different paradigms and programmability of multi-robot systems are presented. The key aspects of cooperative multi-robot systems are discussed along with the different methods in which cooperation is implemented. The use of mobile agents to provide multi-robot system reconfigurability, reprogrammability, and rapid deployment is introduced. Several multi-robot system middleware are discussed along with specialized middleware for cooperative systems.;A highly flexible and reconfigurable cooperative robot control platform called Mobile-R has been developed in the course of this research. Mobile-R consists of two modules: the Robot Control System (RCS) and Deployment System (DS). Mobile-R is a highly extensible platform that follows the multi-agent paradigm. It allows for the implementation of architectures popularly used in the different multi-robot paradigms and is based on widely accepted standards for multi-agent interaction allowing for interoperability with other multi-agent systems. Mobile-R is built upon Mobile-C, an IEEE Foundation for Intelligent Physical Agents standards compliant mobile agent system. The innate mobility characteristic of mobile agents provides an invariant execution of control code over disparate hosts and overall system fault tolerance. The system has been validated through multiple experiments presented in the dissertation. The simulated application of Mobile-R to tier-scalable planetary reconnaissance demonstrates the feasibility and applicability of the system to various multi-robot scenarios.