论文部分内容阅读
控制问题——即在问题求解过程中,一个AI系统如何决定它要执行的动作——是每个智能系统都要解决的问题。本文介绍了智能控制应实现的八点行为目标及实现该目标的黑板控制结构。这种结构把控制问题当作一个实时规划问题,显式地加以解决,具有高度的灵活性。本文采用一个多任务规划的例子来帮助说明黑板控制结构中的一些概念。最后对黑板控制结构的长处与不足作了一个简略的总结。
Controlling problems - that is, how an AI system decides what action it is to perform in solving a problem - is a problem that every intelligent system has to solve. This article describes the eight-point behavioral goals that intelligent controls should achieve and the blackboard control structure that accomplishes that goal. This kind of structure takes the control problem as a real-time planning problem and resolves it explicitly with a high degree of flexibility. This article uses a multitasking example to help illustrate some of the concepts in the blackboard control structure. Finally, a summary of the strengths and weaknesses of the blackboard control structure is given.