改进的免疫遗传算法在多机器人协作中的应用

Application of Artificial Neural Network and Immune-Genetic Algorithm with Elitist to Cooperative Transport of Multi-robots System

针对群体机器人的协作行为,采用一种改进的具有精英保留策略的免疫遗传算法与神经网络相结合的控制策略,实现多移动机器人对单一物体的搬运行为;这种改进的免疫遗传算法,简称为IGAE(Immune Genetic Algorithm with Elitism),结合了精英保留策略(elitism strategy),并将抗体相似度、期望繁殖率以及克隆选择概率采用新的定义方法和计算公式。将人工神经网络作为机器人的行为控制器,并采用改进的具有精英保留策略的免疫遗传算法对神经网络的连接权值进行优化调整,使神经网络可为群体机器人生成最佳的行为决策,从而实现多移动机器人在固定环境中对目标物体的成功搬运。在MATALB环境下的动态仿真实验结果证明了该方法在整个搬运期间具有良好的决策能力。

In this paper, we have developed an innovative control strategy, based on the evolvable artificial neural network and the artificial immune algorithm with elitism for the cooperation bahavior of the Multi--Robot Systems. Using this system we executed performance simulation and parametric design of the cooperative transport behavior for Multi--Robot Systems with single prey. The innovative Immune-- Genetic Algorithm with elitist give the new definitions of parameters for algorithm. This system used a Back Propagation Neural Network with three layers to process the perceptive information of every robot and made a decision for its action. The IGAE was used to optimize the connection weight values of this neural network, which made the performance of the neural network to be evolved continuously and finally a behavior decision--making system with good performance could be obtained. The results of simulation experiments indicate that the mobile robots can keep stable transport of the prey, which proves the good decision--making ability of the IGAE--ANN behavior decision--making system during the transport.