CGF战场空间认知行为建模研究

The purpose of research on CGF behavior modeling is to build a cognitive model which can truly reflect the human decision making process in battle field. In military operations, high level cognition, reasoning and decision-making over battlefield space are mostly carried out in terms of qualitative thinking. Currently in CGF applications, however, the method which used to model human spatial reasoning are simply quantitative, this is far from meeting the requirements of reality and real time response. This program aims to research the modeling theories and methods on CGF-oriented battlefield spatial reasoning. The research is based on the existing achievement of our group on CGF behavior modeling. Taking spatial cognition and reasoning tasks adopted by tactical units as application background, the work would integrate theories and key techniques from Qualitative Spatial Reasoning (QSR), so as to construct CGF spatial cognition mdoels which can perform high level spatil reasoning qualitatively and execute these tasks quantitatively. The main tasks of our research includes: 1) Dynamic detection and local repairmen of battle field topologies; 2) Dynamic planning and reasoning on CGF's tactical spatial constraints networks; 3) Parallelization and multi-layering of tactical spatial reasoning; and 4) Design of Integrated spatial cognition structure model for tactical CGFs. The success of this program will provide novel theories and algorithms for building real, efficient CGF models.

构建真实反映人类战场决策过程的认知模型是计算机生成兵力（Computer Generated Force,CGF）建模理论研究的重要目标。在作战行动中，人对战场空间的高层认知和决策过程大多属于定性思维范畴，而目前CGF空间认知建模中单纯采用的定量描述方法难以满足模型真实性和实时性需求。本项目拟在前期CGF行为建模研究成果的基础上，以战术级作战仿真中兵力实体空间认知建模为应用背景，结合定性空间表示及推理的相关理论和关键技术，开展面向CGF的战场空间认知行为建模研究，力求实现更符合人类认知规律的全局定性、局部定量CGF空间认知模型。主要研究内容包括：1）动态环境下战场空间拓扑结构信息的提取与局部重构；2）动态环境下CGF战术意图空间约束网的规划与求解；3）战术空间推理的分层规划和并行化；4) 战术级CGF 的一体化空间认知体系结构。为构建更为逼真、高效的空间认知行为模型提供新的理论和方法支撑。

构建真实反映人类战场决策过程的认知模型是计算机生成兵力（Computer Generated Force,CGF）建模理论研究的重要目标。本项目以战术级作战仿真中兵力实体的空间认知为应用背景，结合定性空间表示及推理的相关理论和关键技术，重点开展：1）动态环境下战场空间拓扑结构信息的提取与局部重构；2）动态环境下CGF战术意图空间约束网的规划与求解；3）战术空间推理的分层规划和并行化；4) 战术级CGF 的一体化空间认知体系结构建模等方面的前沿研究，取得一系列实践性较强的应用成果，包括：1） 高效的战场地形预处理算法、2）在预处理数据上的微秒级别的实时路径规划算法、3）自适应战术位置学测策略学习模型等一系列核心算法和学习模型。本课题所取得的成果能够有效支撑未来分队作战仿真系统中单兵、群组、作战平台的实时空间感知与推理决策行为建模需求，为未来实现行为模式逼真拟人、算法性能实时高效的分队级虚拟兵力行为模拟提供有效的技术手段支撑。