基于对象模型的机器人规划生成系统及并行推理

The System of Plan Generation and Parallel Reference Based on Object Model for Robot

提出了一种机器人规划生成系统的结构，该系统对机器人物理世界进行了模型化，建立了对象模型知识库，在控制机器人动作序列的生成中，模仿了人的思维方式，将最基本、最简单的最先处理，采用差异减小法逐步减小当前状态与目标状态之间的差距而逼近在达到目标状态，并提出了一种能较好解决画面问题的方法，用该方法解决画面问题比较完美。此外，提出了一种将并行推理技术应用于机器人任务规划的方法，并设计了推理程序。

This paper introduces a new structure of plan generation for robot. In the system,the knowledge hases of theobject are estahlished by modeling the YGR-1 robot and state of the robot. In designing the task scheduling for therobot,we mimicke the made of thinking of man. The simplest and primitive task is processed firstly to reduce increasingly the difference between current state and target state up to target state is achieved. Once more,we propose amethod with which the frame-problem is solved well. ln addition we apply the technologies of parallel inference to plangeneration for the robot and develop the program of parallel inference.