摘要
针对环境未知、目标位置已知情况下机器人路径规划问题,提出了一种基于认知结构SOAR的探索方法。首先,结合箱子移动问题介绍了SOAR的基本概念及求解机制;其次,在讨论了人的行为模型和SOAR的决策过程基础上,设计了基于SOAR-Agent的行为建模方法;最后,利用本文提出的方法,设计长期知识,使机器人能够模拟人在未知动态环境下向目标搜索的行为。通过仿真实验,对比分析了长期知识的完备程度对机器人顺利完成任务的影响,同时也验证了该方法的有效性。
In a situation that environment was unknown and target position was already known, a search method based on cognitive structure SOAR was proposed. The theory and problem solving mechanism of SOAR were discussed with the box moving problem, behavior model of human and decision process of SOAR were analyzed, and behavior modeling method was designed based on SOAR-Agent. By applying the proposed method, robot could simulate the searching pattern of hu. man in a unknown and dynamic environment, the effects of completeness of long term memory on the task of navigating a robot to the destination were analyzed through simulating experiment, and the effectiveness of this method was also verified.
作者
穆杨
MU Yang(Naval Avation University, Huludao Liaoning 125000, China)
出处
《海军航空工程学院学报》
2019年第3期297-302,共6页
Journal of Naval Aeronautical and Astronautical University