Scholarlink

Automated Planning is a well-established field of AI that, although effective, still struggles to gain widespread adoption in industrial settings due to limited ease of use, model maintenance, and performance tuning. Initiatives such as ROSPlan and Unified Planning aim to reduce the gap between planning and robotic applications, but significant expertise is required to design effective plan-based controllers. We investigate the use of a cognitive architecture to integrate heterogeneous planning frameworks and dynamically set decision-making skills. The paper introduces the architecture’s general features and discusses the representation and reasoning capabilities in a mobile manipulation scenario.

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

Towards Reconfigurability of Plan-Based Controllers Through Metacognition

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Alessandro Umbrico,
  • Sebastian Stock,
  • Martin Atzmueller,
  • Amedeo Cesta,
  • Elisa Foderaro,
  • Joachim Hertzberg,
  • Oscar Lima,
  • Juan Carlos Saborío,
  • Marc Vinci,
  • Nicola Pedrocchi,
  • Andrea Orlandini

摘要

Automated Planning is a well-established field of AI that, although effective, still struggles to gain widespread adoption in industrial settings due to limited ease of use, model maintenance, and performance tuning. Initiatives such as ROSPlan and Unified Planning aim to reduce the gap between planning and robotic applications, but significant expertise is required to design effective plan-based controllers. We investigate the use of a cognitive architecture to integrate heterogeneous planning frameworks and dynamically set decision-making skills. The paper introduces the architecture’s general features and discusses the representation and reasoning capabilities in a mobile manipulation scenario.