Abstract

Modular self-reconfigurable robots (MSR) are robots composed of modules that translate over one another to permit reconfiguration and locomotion. This construction allows them to traverse a broader range of environments than legged or wheeled robots. MSR also posses the ability to split apart to parallelize their efforts or combine with each other to produce a larger robot capable of navigating more diverse terrain. In this paper we discuss a state-based reactive architecture for the distributed control of cooperative MSR teams in unknown environments. The MSR use local sensory data from the environment and a model of the team to select their actions. These actions include selecting a destination, aborting a route to a destination, splitting into two separate robots, and combining with another robot. In simulation, team-configuration, environmental complexity, and behavioral parameters are varied to discern the most effective circumstances for the architecture and MSR. Our results show that the best configuration of the system is highly dependent on the environment.

Library of Congress Subject Headings

Robots--Control systems--Design and construction; Robots--Kinematics; Genetic programming (Computer science)

Publication Date

2009

Document Type

Thesis

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Butler, Zack

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: TJ211.35 .F32 2009

Campus

RIT – Main Campus

Download

Share

COinS