Robots with a Sense of Self: Exploring Embodied Intelligence
摘要
This chapter explores the concept of embodied intelligence, emphasizing the integration of physical interaction, perception, and cognition in artificial agents. It begins by outlining foundational theories that position embodiment as central to intelligent behaviour, contrasting it with disembodied computational models. The chapter then introduces embodied cognitive agents—systems that perceive, act, and learn through interaction with their environments—and examines how embodiment shapes learning and decision-making processes. It also discusses the development of theory of mind and communication in embodied agents, highlighting their capacity to infer intentions and engage in social interaction. Advanced topics include multi-agent collaboration, sensorimotor learning, and the challenges of scaling embodied systems to complex, real-world tasks. Together, these sections provide an understanding of how physical embodiment enhances adaptability, autonomy, and cognitive richness in artificial intelligence.