Research on Design of Dynamic Collaborative Systems by Organizing Perceptual-Action Concepts within Human-Machine Coexisting Environments

在本研究项目中，我们通过在人机共存环境中组织感知-行动概念，开发了一个动态协作系统。这个项目的主要目的是让机器系统能够与人类（即工厂的用户和操作员）互动，并通过观察他们的行为来捕捉他们的概念，通过这种能力，机器系统将能够成为人类的友好伙伴，通过交替他们的工作，并为他们提供自己的观点来指导他们，有时批评人类的行为。这种人与机器代理之间的持续互动将与人类建立一种新的伙伴关系，而不是传统的固定角色分配给人与机器。在这个项目中，我们着重于机器的概念学习，包括概念形成、动态模型构建、运动概念共享等。特别介绍了遗传算法和遗传规划等进化计算技术。在此基础上，我们将我们的想法应用于界面代理做出高风险、时间关键的决策。在第二年，我们分析了人类和这种人工智能系统产生的交互，以及它们的动态和紧急属性。我们处理的目标域是；以人为中心的自动化接口代理，以及移动机器人的远程操作系统。

In this research project, we developed a dynamic collaborative systems by organizing perceptual-action concepts within the human-machine coexisting environments. The main purpose of this project is to let the machine system to be able to interact with the human (i.e., users and operators of the plants) and to capture their concepts by observing their behaviors, By this capability the machine system would be able to become a friendly partner for the human by alternating their works and by providing them with its own views for guiding them and sometimes for criticizing the human behaviors. Such continuous interactions between the human and the machine agents will establish a new partnership with the human instead of the conventional fixed roles allocated to the human and the machine.In this project, we focused on the conceptual leaming of the machine including concept formation, dynamic model construction and sharing motion concepts, etc. Especially, evolutional computation techniques such as genetic algorithms and genetic programming are introduced for this purpose. Based on that, we applied our ideas to an interface agent's making high-stakes, time-critical decisions, In the second year, the interactions generated by the human and such artifact agent systems were analyzed wit h respect to their dynamics and emergent properties. The target domains we dealt with are ; interface agent for human-centered automation, and a teleoperation system for mobile robots.