A developmental genetics-inspired approach to robot control
A developmental genetics-inspired approach to robot control
复制标题
一种受发育遗传学启发的机器人控制方法
DOI：
10.1145/1102256.1102325
复制
发表时间：
2005
影响因子：
9.2
通讯作者：
Sanjeev Kumar
中科院分区：
文献类型：
--
作者：
Sanjeev Kumar
构建模块化的、可扩展的、能够适应、自组装和自我修复的复杂技术的需求，激发了人们对使用受发育生物学启发的方法的新兴趣。为了满足这一需求，出现了一个新的领域，称为计算开发（CD）。它的重点是适应发育生物学的过程和机制，以帮助我们建立可扩展的复杂技术。然而，由于该领域的萌芽性质，研究这些方法在不同问题领域的潜力对其成功至关重要。本文探讨了将发育生物学启发的方法应用于反应性机器人控制这一要求苛刻的问题领域的可行性。利用发育遗传学作为灵感来源，将遗传调控网络模型与空间分布式进化算法结合使用，以进化实时机器人控制器来完成诸如通用避障等任务。
The need to build modular, scalable, and complex technology capable of adaptation, self-assembly, and self-repair has fuelled renewed interest in using approaches inspired by developmental biology. To meet this need, a new field, called Computational Development (CD), has emerged. Its focus is on adapting processes and mechanisms from developmental biology so as to help us build scalable, complex technology. Due to the embryonic nature of the field, however, research investigating the potential of such approaches for different problem domains is crucial to its success. In this paper, the plausibility of applying a developmental biology-inspired approach to the demanding problem domain of reactive robot control is explored. Using developmental genetics as a source of inspiration, a model of genetic regulatory networks is used in conjunction with a spatially distributed evolutionary algorithm to evolve real-time robot controllers for tasks such as general purpose obstacle avoidance.