Active Observer

9307126 Bajcsy这是三年持续奖的第一年。本研究涉及主动观察者视觉行为的建模、分析和综合研究。活动观察者是一种移动的代理，其任务是观察和报告/通信关于另一个代理的活动的信息，其任务可能是操作和/或导航。主动观察者是一个以合作方式共同工作的小代理团队的一部分。团队中的所有代理都配备了视觉导航功能，但他们所从事的任务不同，这取决于他们在团队中的角色和目标。代理可以独立导航或以合作方式导航（例如，在两个操纵器代理一起携带对象的情况下）。对于以合作方式导航的操纵代理，通信问题就出现了；因此，如何将视觉信息表示为交流的问题就出现了。观察者-代理的视觉任务将根据他们的观察是否服务于其他代理的导航或操作目的而有所不同。本研究的主要重点是观察者必须拥有的控制结构，以便为手头的任务提供适当的观察。为了对行为进行建模，采用了离散事件系统（DES）理论中的一种形式，适合研究系统的控制理论问题。预计本研究的贡献将在：1)建立视觉引导行为模型，将视觉处理与控制结构紧密耦合；2)识别导航和通信所需的基本行为及其组成，从而导致更复杂的行为。两种基本行为被区分开来，一种是观察与物理行为直接相关，另一种是观察和行为要么被接收，要么被传递。DES形式主义的使用允许以系统的方式合成复杂的行为，并保证它们的可控性和可观察性

9307126 Bajcsy This is the first year of a three year continuing award. The research involves investigation of modeling, analysis and synthesis of visual behaviors of Active Observers. An Active Observer is a mobile agent whose task is to observe, and report/communicate information about the activity of another agent(s) whose task may be manipulation and/or navigation. Active Observers are part of a small team of agents who work together in a cooperative fashion. All agents in the team are equipped with visual navigation capabilities, but the tasks they are engaged in differ, depending on their role in the team and their goal. The agents could navigate independently or in a cooperative fashion (i.e. in the case of two manipulator agents carrying an object together). For manipulatory-agents navigating in a cooperative fashion, the issue of communication arises; hence, the problem of representing visual information for communication. The visual tasks of the observers-agents will differ depending on whether their observations serve for the navigation or manipulation purposes of the other agents. The main emphasis in this research is on the control structure that the observer must have in order to deliver the proper observations for the task at hand. For the purpose of modeling behaviors a formalism from the Discrete Events Systems (DES) theory has been adopted, suitable for investigating control-theoretic issues of a system. It is anticipated that the contribution of this research will be in: 1) modeling visually guided behaviors which tightly couple visual processing with control architecture; and 2) the identification of elementary behaviors needed for navigation and communication and their composition leading to more complex behaviors. Two kinds of elementary behaviors are distinguished, one where observations are directly connected with physical actions, and the other where observations and actions are either received or transmitted. The use of DES formalism allows the synthesis of complex behaviors in a systematic fashion and guarantees their controllability and observability