基于遗传神经网络的虚拟履带车辆自主导航

Auto-navigating virtual tracklayer based on genetic artificial neuron network

为了解决计算机生成兵力CGF(com puter generated forces)实体在虚拟环境中自主导航、进行路径规划的基本问题,以虚拟履带车辆为例,提出了采用神经网络模型对其在三维地形中的自主导航智能行为进行建模。采用遗传算法作为神经网络进行无导师学习的算法,通过仿真实验对模型的可行性和有效性进行验证。遗传神经网络模型经过若干代进化能使虚拟车辆得到较好的行进路径。实验结果证明了模型和算法的正确性。

To solve the basic problem in the auto-navigating and course programming, with the virtual tracklayer as an example, the auto-navigating behavior of the tracklayer was analyzed, and a modeling method based on ANN（artificial neural network） was prepossessed to simulate the auto-navigating behavior of the virtual tracklayer in 3D virtual environment. GA （genetic algorithm） was used as the learning method of the ANN model in non-tutor learning mode. The model was validate through a simulation experiment. The results show that the virtual tracklayer can find a reasonable path and get to the target after the evolution of the genetic ANN for some generations. It turns out that the model and the algorithm are reasonable.