NSF-CNPq Collaborative Research: Combining Cognitive & Utilitarian Coordination in a Layered Agent Architecture

该 CNPq 研究项目将使马萨诸塞大学阿默斯特分校的一个研究小组与巴西南里奥格兰德大学人工智能小组的一个研究小组合作。他们将共同创建一个两级多代理框架。在更高的层面上，将开发一个对符号信息进行推理的子系统。其活动涉及特定领域的协调任务。在较低级别，将开发一个对定量信息进行推理的子系统。其活动涉及可行性和实施操作。这两层包含异步操作的相互依赖的活动。下层向上层提供反馈；上层通过在必要时改变其任务或绩效目标来响应反馈。这样的架构将使多代理系统更容易在开放环境中运行，例如互联网，其中要实现的目标、可用资源和其他交互代理都是动态变化的。

This CNPq research project will enable a research group at the University of Massachusetts, Amherst to cooperate with a research group at the Artificial Intelligence Group at the University of Rio Grande do Sul in Brazil. Together they will create a two-level multi-agent framework. At the higher level, a subsystem that reasons about symbolic information will be developed. Its activities pertain to domain-specific coordination tasks. At the lower level, a subsystem will be developed that reasons about quantitative information. Its activities pertain to feasibility and implementation operations. The two layers comprise interdependent activities that operate asynchronously. The lower layer provides feedback to the upper layer; the upper layer responds to the feedback by altering its tasks or performance objectives where necessary. Such an architecture will make it easier for multi-agent systems to function in open environments, such as the Internet, where the goals to be achieved, the resources available and the other interacting agents are dynamically changing.