Research On Distributed Control Based On Nonideal Cellular Automata For Self-reconfigurable Robot

Self-reconfigurable robot consists of multiple connected modules, and and thenumerous DOFs of the modules has aggravated the difficulty for its motionplanning. Traditional centralized motion control focused on the reconfigurationfrom initial configuration to some certain configuration, which has been proved tobe a NP complete problem, leading to an exponent rising of planningcomplexity.Distributed control motion solved all the problems above by lettingevery module carry out the action planning work itself, which makes the best use ofeach modules’ computing resource and increases the robot’s self-repair ability andadaptability to the environment.The research on cellular automation theory in self-reconfigurable robot aremainly concentrated in ideal cellular automata which use the kinematic cubic as theideal cell.The difference between real module and ideal cell in mobility restrict theuse of cellular automation theory in the field of self-reconfigurable robot.Based on the research of ideal plane cellular automata, we use the cube as theideal cell module, and build the ideal3D cellular automata for self-reconfigurablerobot. The ideal cell can move on the face of other cells freely. We research into theinteraction pattern between cell mobility and cell rules, choose limited cell actionsto form the ideal cell rules. Which enables the reconfiguration locomotion ofself-reconfigurable robot over uneven terrain.In order to build the nonideal cellular automata for actual robot system, weanalyze how actual robot modules gather environment information and executemechanical motion. Take module motion limitation into account, we use the metamodule as independent cell for robot system. Four kinds of UBotmeta modulediffers in numbers of modules and relative position of connected modules areproposed. Motion features of meta module in reconfiguration locomotion arecompared from each other in multiple abstracts, and only one meta module ischosen as the best ideal cell model for UBot self-reconfigurable robot.The interaction form between nonideal cell model and homologous cell ruleshas been analyzed to build nonideal cellular automata model for UBotself-reconfigurable robot. Distributed controller for UBot module has beendesigned, and implemented in each module. The distributed control ofself-reconfigurable robot has been realized. Self-reconfigurable robot can adapt unknown environment throughreconfiguration locomotion. Distributed control reconfiguration locomotion avoidsthe complexity of those environment which fixed configuration motion can hardlyget through, increasing its usage in realistic conditions. The modules handle t heenvironment information independently, and each modules’s mobility lead to theregular distributed reconfiguration locomotion for robot system. The validities ofthe distributed control strategy and the nonideal cellular automata model areverified by the simulation in environments with different complex obstaclesSelf-reconfigurable robot can emerge new special configuration according tomission requirement, which is its main advantage over traditional robot and the corevalue of self-reconfigurable robot. Taking the environment and mission request asthe guidance for self-reconfiguration, the problem of configuration has beenrenewed. Theories from commercial manufacture and natural selection are adoptedto help the self-reconfigurable robot accomplishs reconfiguration in the process ofenvironment adaption. The distributed self-reconfiguration of UBotself-reconfigurable robot is realized.