上海大学学报(社会科学版) ›› 2026, Vol. 43 ›› Issue (1): 86-96.

• • 上一篇    下一篇

具身人工智能的身体构建

  

  1. 北京师范大学 哲学学院
  • 出版日期:2026-01-15 发布日期:2026-02-12
  • 作者简介:董春雨,北京师范大学哲学学院教授。研究方向:物理哲学及复杂性哲学。
    王逸凡,北京师范大学哲学学院博士研究生。研究方向:科学技术哲学。
  • 基金资助:
    国家社会科学基金重大项目(24&ZD320)

The Construction of Embodied AI Bodies

  1. School of Philosophy, Beijing Normal University
  • Online:2026-01-15 Published:2026-02-12

摘要: 具身人工智能的具身性内涵与人类具身认知存在差异:人类以天然生物身体的感知运动系统为认知根基,形成包含意识、情感与行动的有机生命整体;而机器则体现为物理载体与算法模型的功能性耦合。对比人与机器的身心关系可知,人工智能虽无法复制人类具身认知的生命属性,却能在功能层面模拟重构“身心关系”,形成“感知-行动”的人工回路。当前机器身体的构建面临三重挑战:形态复杂性问题、控制复杂性问题和可进化性问题。未来研究可以聚焦于这几个基本问题:在机器身体、机器信息处理系统与外部环境之间的平衡交互的基础上明确若干机器身体构建的方向。

关键词: 具身人工智能, 具身性, 身心关系, 感知-行动回路

Abstract: Embodiment in Artificial Intelligence (AI) fundamentally differs from human embodied cognition. Human cognition is grounded in the sensory-motor system of a biological body, forming a holistic entity that integrates consciousness, emotion, and action. In contrast, machine embodiment is achieved through the functional coupling of a physical carrier and an algorithmic model. A comparative analysis reveals that while AI cannot replicate the biological properties inherent to human cognition, it can simulate a functional“mind- body relationship”by creating an artificial perception-action loop. Nevertheless, the construction of effective machine bodies faces three primary challenges: morphological complexity, control complexity, and evolvability. Future research should therefore focus on resolving these fundamental issues by investigating the balanced interaction among the machine’s body, its information processing system, and the external environment to chart a clear path forward for machine-body engineering.

Key words: embodied AI, embodiment, mind-body relationship, perception-action loop

中图分类号: