摘要
为了解决智能仓储系统中自动导引车(Automated Guided Vehicle,AGV)的全局路径规划问题,提出了一种基于改进Q-learning算法的AGV全局路径优化算法。针对运用传统Q-learning算法只适用于离散状态模型或者较为简单的连续状态模型的情况,将人工神经网络引入Q-learning算法。通过仿真实验进行对比,仿真结果表明,新算法可以使AGV通过自主学习自行寻找出一条无碰撞最优路径,证明了改进后的Q-learning算法的可行性。
In order to solve the problem of AGV global path planning in intelligent storage system,this paper proposes an AGV global path optimization algorithm based on improved Q-learning algorithm.Aiming at the situation that the traditional Q-learning algorithm is only suitable for discrete state model or relatively simple continuous state model,this paper introduces artificial neural network into Q-learning algorithm.Through the comparison of simulation experiments,the simulation results show that the new algorithm can make the AGV find an optimal path without collision by itself through independent learning,which proves the feasibility of the improved Q-learning algorithm.
作者
王鼎新
WANG Dingxin(School of Automation,Qingdao University,Qingdao 266071,China)
出处
《电子设计工程》
2021年第4期7-10,15,共5页
Electronic Design Engineering
基金
国家自然科学基金(61573205)
山东省自然科学基金(ZR2015FM017)。
