Research On Distributed Reconfiguring Strategy Of Modular Self-Reconfigurable Robots

Modular self-reconfigurable(MSR)robots are consisting of multi basic modules.Each module is a complete robot with independent sensing,moving ability and onboard battery.Modules can connect with directly neighboring modules in diverse relative orientations.The main novelty of MSR robots lies in the ability to change global structure according to current environments and tasks at hand.In decentralized control,modules do motion planning only by environmental information in limited sensible range and communication with directly connected modules.For the absence of information about global state,there are still many corner stones on the way to physical implementation of MSR robots,such as the convergence and global connectivity.The decentralized reconfiguration locomotion(surface locomotion)for global movement and decentralized selfreconfiguration for changing global structure are still two open problems in the research of MSR robots.The decentralized reconfiguration locomotion relies on continuous motion of independent modules from back to front on the surface of other modules.Robots can get through obstacle environments by local motion planning of independent modules.Existing research has shown the implementing potential of Cellular Automata(CA)on the control of decentralized reconfiguration locomotion.But the design of CA rules depends greatly on personal experience of designers and different rule sets are needed according to obstacle states in the environment,which limits the self-adaption of MSR robots to open environments.From the basic motion of modules,a simplified set of CA rules are designed in this work.This new set of rules is greatly simplified in rule numbers compared with rules in existing work.And the problems of global connectivity and local collision by different modules moving to the same position are also solved through corresponding distributed manners.Simulations and experiments are provided to illustrate the function of designed rules and proposed methods.The decentralized self-reconfiguration of MSR robots makes full use of computation and motion ability of distributed modules.The decentralized reconfiguration strategy solves the exponential increase of planning complexity in centralized control.For the absence of global state,distributed modules move independently and simultaneously according to limited accessible information.Both the design of target structure and the convergence of reconfiguration process are still open to researchers in this field.We turn to the nature for inspiration and introduce the fractal theory into the decentralized self-reconfiguration of MSR robots.For the high ability to construct complex structures using simple symbols,L-systems are used to design target configuration for MSR robots through rewriting.The turtle interpretation in computer graphics is extended to MSR robots.Simulations are done to prove the effectiveness of the proposed strategy in convergence and scalability.Main influence factors on convergent process are also analyzed.MSR robots working in open environments need to reconfigure continuously in different patterns,such as locomotion through narrow space and then reconfigure to some configuration for global function.Strategies for controlling the continuous translation among different patterns are strongly needed for implementation reasons.Inspired by the common adaption of natural plants,parametric L-systems are introduced and extended for the adaption control of MSR robots.Parametric rewriting rules are designed for bringing environmental influences into the reconfiguration process.MSR robots can change reconfiguring state from the current state directly by the proposed strategy.To verify the effectiveness of the proposed strategy for decentralized reconfiguration locomotion and self-reconfiguration,multi experiments on physical Seremo robots are done.Seremo robots are designed as modular self-reconfigurable mobile robots.Experiments with 4 and 8 modules are done in indifferent obstacle environments to illustrate the function of designed CA rules and the proposed strategy for global connectivity and local collision problem.A hybrid reconfiguration strategy by combining CA and L-systems are implemented on Seremo robots with different target configurations to verify the convergence and scalability of distributed self-reconfiguration.A Seremo robot with 9 modules is also used to reconfigure between cross-shape structure and L-shape structure,as well as from reconfiguration locomotion to self-reconfiguration with a 6 shape structure as target configuration.These experiments illustrate the effectiveness of the proposed strategy for continuous reconfiguration.