Analysis and Control of Behavior of Insect by Artificial Neural Network

随着观测和测量方法的进步，昆虫行为的定性特征和产生昆虫运动的外骨骼的精确结构越来越清楚。另一方面，对行为产生的中枢神经系统及其控制机制的研究也在积极进行。本研究在具有学习过程的移动机器人控制系统中，先后构建了行为涌现的神经机制。选择蚕蛾的神经系统作为模型系统。蚕蛾的神经细胞数量少，非常适合作为模型。长期以来，Kanzaki从神经生理学、组织学和行为科学的角度对行为产生的中枢神经机制及其控制系统进行了研究。他尤其对雄性蚕蛾感知雌蛾信息素接近雌蛾时的固定行为（之字形和旋转运动）感兴趣。Miura和Shimoyama开发了一种轮式移动机器人，用于追踪雌性蛾的信息素。在这个机器人中，从真正的雄性飞蛾的天线上切下两个非常小的部分（5mm长）作为信息素的传感器。通过通用算法，机器人学习了蚕蛾的上述固定行为。

The qualitative characteristics of the behavior of insects and the precise structure of the external skeleton which generates the motion of insects are getting clearer by the progress of the method of observation and measurement. On the other hand, the research on the central neural system in the emergence of the behavior and the control mechanism of it is also being conducted actively.In this research, the neural mechanism of the behavior emergence was successively constructed in the control system of the mobile robot with learning process. As the model system, the neural system of the silk worm moth was selected. The number of neural cells of the silk worm moth is small and very suitable as the model.Kanzaki has conducted research on the central neuro-mechanism for the behavior emergence and its control system from standing point of the neurophysiology, the histology and the behavioral science for long time. Especially he was much interested in the fixed behavior (zig zag and rotation motion) of the male silk worm moth to approach the female moth sensing female moth pheromone.Miura and Shimoyama have developed the wheeled mobile robot which pursue the female moth pheromone. In this robot, two very small parts (5mm length) of the antenna cut from the real male moth are used as sensors of pheromone. By generic algorithm, the robot learned the above mentioned fixed behavior of the silk worm moth.