基于仿生机器鱼的人工生命模型

Artificial Life Model Based on Bionic Fish

为了构建具有自学习、自适应、自协调的人工生命体,结合对鱼类生理结构和行为机理的研究和分析,提出了基于人工生命的仿生机器鱼系统。将非接触式供电技术应用于水下供电,模拟鱼类生理活动,在仿生鱼上实现了动态、无连接、不间断的电能供给,并且基于鱼类行为的层次结构,实现了"动机-行为-动作"的3层行为决策系统,引入强化学习算法,模拟鱼类与环境交互的后天学习过程。通过实物仿生鱼的游姿优化实验,证明了研究和设计工作的有效性。

In order to build self-learning, adaptive, self-coordination of intelligent agent, and combining with the research and analysis on physiological structure and behavior mechanism of natural fish, we proposed the artificial life system model of bionic fish. The paper applied non-contact power supply technology in underwater electrical supply, simulated physiological activity of natural fish, and real- ized dynamic, no-connection, uninterrupted power supply in bionic fish; Based on fish behavior hierarchy, the paper achieved the ＂motive-behavior-action＂ three layers decision-making system, and introduced the reinforcement learning algorithm to simulate environmental interaction learning process. The experiment of swim posture optimization verified the effectiveness of the research and design work.