...
首页> 外文期刊>Journal of energy and power engineering >Use of Context Blocks in Genetic Programming for Evolution of Robot Morphology
【24h】

Use of Context Blocks in Genetic Programming for Evolution of Robot Morphology

机译:在遗传编程中使用上下文块促进机器人形态的发展

获取原文
获取原文并翻译 | 示例
           

摘要

The paper explores applications of genetic programming to co-evolution of morphology and low-level control. In most reasonably difficult tasks, facilitation provided by modularity has proved to be vital for successful application of genetic programming. However, the need for sharing data among nodes in the syntactic tree becomes especially acute when evolving modular programs. It has been shown before that it may be beneficial that modules themselves be node-attached. The paper presents extensions to standard genetic programming (the so-called contexts and context blocks) that allow for straight-forward storage, retrieval, transfer, and modification of data stored in the context of a syntactic tree, and shared by multiple nodes. Framework is thus provided for both: data sharing and node-attached modules. Finally, using context blocks, a genetic algorithm has been embedded within genetic programming to evolve values of constants. In genetic programming evolution of constants has been a long-standing problem. The paper shows how context blocks can be utilized to provide a more granular and flexible approach to their evolution. As shown in previous works, node-attached modules perform favorably when compared with existing approaches. Concerning evolution of context block constants, it is shown here that they too perform favorably when compared with ephemeral constants.
机译:本文探讨了遗传编程在形态学和低水平控制共同进化中的应用。在最合理的困难任务中,模块化提供的便利对成功应用基因编程至关重要。但是,当开发模块化程序时,语法树中的节点之间共享数据的需求变得尤为迫切。之前已经表明,将模块自身连接到节点可能是有益的。本文介绍了对标准遗传程序设计的扩展(所谓的上下文和上下文块),这些扩展允许对存储在语法树上下文中并由多个节点共享的数据进行直接存储,检索,传输和修改。因此,提供了同时用于数据共享和节点连接模块的框架。最后,使用上下文块,遗传算法已嵌入遗传编程中以演化常数的值。在遗传编程中,常数的演化是一个长期存在的问题。本文展示了如何利用上下文块为它们的演化提供更细化和灵活的方法。如先前的工作所示,与现有方法相比,连接节点的模块性能良好。关于上下文块常量的演变,此处显示它们与临时常量相比也表现出色。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号