针对涉身行为的自我意识机器人构建方法及其实现

项目名称： 针对涉身行为的自我意识机器人构建方法及其实现

项目编号： No.61273338

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 自动化技术、计算机技术

项目作者： 周昌乐

作者单位： 厦门大学

项目金额： 80万元

中文摘要： 机器意识可以为机器人及信息技术的进步提供新的契机，因此近年来，机器意识的研究工作，引起了国际学术界的广泛关注。在这其中，一个关键又困难的课题，就是自我意识的机器实现。在这方面，以往的研究，不是将自我意识看作是一种元认知能力，就是将自我意识误解为一种"自我"的意识，很少涉及真正体现其内涵的内省反思能力。因此，本课题旨在针对诸如感知、言语、行动等涉身行为，着眼实现内省反思能力，来构建一种自我意识计算方法，并应用到智能机器人系统的实现之中。具体的研究是通过给出涉身行为的内部表征，利用神经-符号集成计算方法，建立体现内省反思机制的扫视、觉知、学习与模拟等功能模块，搭建一种自我意识的计算体系结构。然后将其植入我们原有的仿脑机器人系统之中，最终形成具有一定自我意识能力的机器人系统。显然，这样的研究，对于破解内省反思机制的机器实现难题，丰富自我意识机器人系统新的构建途径，无疑是有着重要的建设性意义。

中文关键词： 机器意识；异步自省理论；自我意识机器人；涉身行为；内省反思计算

英文摘要： Machine consciousness is expected to create new opportunities in robotics and information technology. Recently, more and more people focus on research work of machine consciousness for academic goals. One of the essential and difficult topics is how to realize self-consciousness by machine. Unfortunately, as far as current researches on computational models of self-consciousness are concerned, researchers often regard self-consciousness as meta-cognitive ability or consciousness of the "self", rather than introspective and reflective capabilities of the machine, which are real connotations of self-consciousness. In order to develop the computational method of self-consciousness as introspective and reflective capabilities, we present a proposal as to how to create a self-conscious robot system by focusing on embodied behaviors, such as perception, speech, and action. The project will utilize neural-symbolic integration computation to build introspective and reflective function modules including introspect saccade, behavior awareness, behavior learning and behavior simulation based on the representation of embodied behaviors. Then we will provide a computational architecture for a self-conscious robot with all introspective and reflective function modules, and set in the brain-like robot system which we have deve

英文关键词： Machine consciousness；Asynchronous introspection theory；Self-conscious robot；Embodied behavior；Introspective and reflective computation