Scholarlink

<p>This paper demonstrates the evolution and extraction approach for the controller of the robotic swarm. The collective perception task has received a lot of attention in the field of swarm robotics. In addition, recent studies showed that the evolutionary robotics (ER) approach successfully designed decision-making strategies for the task. This study focused on a detailed analysis of the evolved decision-making strategies. As in related work, the artificial neural network (ANN) was employed to approximate a decision-making mechanism. At first, we examined how the available information for ANN affects the performance of the collective perception task. Secondly, a visualization approach was proposed to extract evolved decision-making mechanisms from ANNs. The computer simulations showed that our visualization approach successfully extracted the evolved decision-making mechanisms as reusable or analyzable by any others.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Evolution and extraction of decision-making mechanisms in collective perception of a robotic swarm

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年10月8日

查看全文
  • Daichi Morimoto,
  • Motoaki Hiraga,
  • Kazuhiro Ohkura,
  • Masaharu Munetomo

摘要

This paper demonstrates the evolution and extraction approach for the controller of the robotic swarm. The collective perception task has received a lot of attention in the field of swarm robotics. In addition, recent studies showed that the evolutionary robotics (ER) approach successfully designed decision-making strategies for the task. This study focused on a detailed analysis of the evolved decision-making strategies. As in related work, the artificial neural network (ANN) was employed to approximate a decision-making mechanism. At first, we examined how the available information for ANN affects the performance of the collective perception task. Secondly, a visualization approach was proposed to extract evolved decision-making mechanisms from ANNs. The computer simulations showed that our visualization approach successfully extracted the evolved decision-making mechanisms as reusable or analyzable by any others.