Dynamics Based Intelligence of Redundant Swarm Mechanical Systems with Adaptability

研究了具有自适应性的冗余群机械系统的动态智能问题。动力学对于生物的行动智能是必不可少的。本研究的目的是将基于个体的动态控制概念推广到基于群体的动态控制。我们考虑以下4个主要议题。(1)。考虑了具有可变约束的混合系统：活蛇的鼻窦提升和侧绕运动以及蛇形机器人的跳跃功能。(2)。具有可变结构功能的模块化机器人系统：采用PI2学习方法生成三足模块化机器人单个单元的运动策略。(3)。群体机器人的分散控制：设计了仅利用局部感知信息对多个移动机器人进行分散控制。(4)。蚂蚁的利他行为分析：利用基于PDE模型的模拟器，分析了困扰行为作为利他行为的有效性。

In this study dynamics based intelligence of redundant swarm mechanical systems with adaptability is considered. Dynamics is essential to intelligence of action of living things. The objective of this study is extension of concept of dynamics based control of individual to that of swarm. We consider following 4 main topics. (1). Hybrid system with changeable constraints: Sinus-lifting and side-winding motions of a living snake and a jumping function of a snake-like robot are considered. (2). Modular robot systems with function of changeable structure: Moving strategies of one unit of three-legged modular robot are generated by the learning method PI2. (3). Decentralized control of swarm robots: Decentralized control of multiple mobile robots using only local sensing information is designed. (4). Analysis of altruistic behavior of ants: By using simulators based on a PDE model effectiveness of the plaguing behavior as an altruistic behavior is analyzed.