Abstract: To solve the problems of traditional Genetic Algorithm (GA) in path planning of Automated Guided Vehicle (AGV) such as easy to fall into local optimum,slow convergence and non-shortest path length,an improved GA combining artificial potential field method and simulated annealing idea was proposed.Combined with artificial potential field method,a guided initial population generation strategy was designed to improve the initialization speed of the algorithm.Then,constraint conditions such as the size of the rotation Angle and the number of unnecessary turns were added into the fitness function to improve the smoothness of the path.Based on simulated annealing algorithm,the selection operator was improved to enhance the global search ability.Edit distance was introduced to screen individuals before crossover to prevent invalid crossover,and the delete operator was added to solve the problem of redundant nodes and get a shorter path.The experimental simulation results showed that the improved algorithm had shorter path length and better convergence effect,and effectively prevented the algorithm from falling into local optimum.After the ROS test platform verification,the search path is more advantageous,which proved the effectiveness and feasibility of the improved algorithm to a certain extent.

Key words: genetic algorithms, artificial potential field, simulation degradation algorithm, automatic guided vehicle, path planning

摘要： 针对传统遗传算法在规划自动导引小车路径时易陷入局部最优、收敛慢且路径长度非最短等问题,提出一种融合人工势场法和模拟退火思想的改进遗传算法。首先,结合人工势场法设计了一种引导式初始种群生成策略来提高算法的初始化速度;然后,将转角大小、非必要转向次数等约束条件加入适应度函数提升路径的平滑性,基于模拟退火算法改进选择算子来增强全局搜索能力,引入编辑距离筛选交叉前的个体以防止无效交叉,并添加删除算子解决冗余节点问题,获得了较短路径。最后通过实验仿真表明,改进算法规划的路径较短、收敛效果较好,有效防止了算法陷入局部最优。后经ROS机器人操作平台验证,搜索到的路径更具优势,在一定程度上证明了改进算法的有效性和可行性。

关键词: 遗传算法, 人工势场法, 模拟退化算法, 自动导引车, 路径规划