Genetic Programming-based Self-reconfiguration Planning for Metamorphic Robot  

作　　者：Tarek Ababsa Noureddine Djedi Yves Duthen 

机构地区：[1]LESIA Laboratory,University of Biskra,PO Box 145 RP,Biskra 07000,Algerla [2]IRIT Laboratory,University of Toulouse 1 Capitole,21 Alles de Brinne,Toulouse 31032,France

出　　处：《International Journal of Automation and computing》2018年第4期431-442,共12页国际自动化与计算杂志（英文版）

摘　　要：This paper presents a genetic programming based reconfiguration planner for metamorphic modular robots. Initially used for evolving computer programs that can solve simple problems, genetic programming （GP） has been recently used to handle various kinds of problems in the area of complex systems. This paper details how genetic programming can be used as an automatic programming tool for handling reconfiguration-planning problem. To do so, the GP evolves sequences of basic operations which are required for transforming the robot＇s geometric structure from its initial configuration into the target one while the total number of modules and their connectedness are preserved. The proposed planner is intended for both Crystalline and TeleCube modules which are achieved by cubical compressible units. The target pattern of the modular robot is expressed in quantitative terms of morphogens diffused on the environment. Our work presents a solution for self recontlguration problem with restricted and unrestricted free space available to the robot during reconfiguration, The planner outputs a near optimal explicit sequence of low-level actions that allows modules to move relative to each other in order to form the desired shape.This paper presents a genetic programming based reconfiguration planner for metamorphic modular robots. Initially used for evolving computer programs that can solve simple problems, genetic programming （GP） has been recently used to handle various kinds of problems in the area of complex systems. This paper details how genetic programming can be used as an automatic programming tool for handling reconfiguration-planning problem. To do so, the GP evolves sequences of basic operations which are required for transforming the robot＇s geometric structure from its initial configuration into the target one while the total number of modules and their connectedness are preserved. The proposed planner is intended for both Crystalline and TeleCube modules which are achieved by cubical compressible units. The target pattern of the modular robot is expressed in quantitative terms of morphogens diffused on the environment. Our work presents a solution for self recontlguration problem with restricted and unrestricted free space available to the robot during reconfiguration, The planner outputs a near optimal explicit sequence of low-level actions that allows modules to move relative to each other in order to form the desired shape.

关 键 词：Modular robots unit-compressible modules SELF-RECONFIGURATION genetic programming reconfiguration planning. 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]