摘要
由于3D打印机器人每次从起点到终点的最优或次优路径不止一条,而且在移动过程中要进行避障,加大了路径规划难度。针对上述问题,提出一种基于改进Q学习的3D打印机器人路径生成方法。根据齐次变换原理和机器人的结构参数分析机器人运动学原理,计算3D打印机器人末端在世界坐标系的位姿,明确3D打印机器人各轴联动关系。将机器人内部的信息节点组成分布式导航网络,控制机器人做出独立导航决策,辅助机器人航向选择。利用改进的Q学习方法,将信息决策导航结果与机器人学习的最佳状态及动作匹配,利用回报函数和贪婪策略方法提高其学习率,实现3D打印机器人路径生成。实验结果表明,所提方法路径规划能力强,计算耗时短。
At present,there is more than one optimal or suboptimal path for 3D printing robots from the starting point to the endpoint,and obstacle avoidance during the movement may increase the difficulty of path planning.Therefore,a method of path generation for the 3D printing robot based on improved Q learning was proposed.Based on the principle of homogeneous transformation and the structural parameters of the robot,the kinematics principle of the robot was analyzed.Meanwhile,the position and posture of the end of the 3D printing robot in the world coordinate system were calculated,and then the interactive relationship between the axes of the 3D printing robot was clarified.Moreover,the information nodes of the robot were formed into a distributed navigation network controlling the robot to make independent navigation decisions and assisting in the heading selection.After that,the improved Q learning method was adopted to match the decision-making navigation result with the optimal state and action of robot learning.Finally,the reward function and greedy strategy were used to improve the learning rate,thus achieving the path generation.Experimental results show that the proposed method has stronger path planning ability and less computational time.
作者
洪涛清
高雪芬
HONG Tao-qing;GAO Xue-fen(Faculty of Mathematics and Computer,Lishui University,Lishui Zhejiang 323000,China;School of Science,Zhejiang Sci-Tech University,Hangzhou Zhejiang 310018,China)
出处
《计算机仿真》
北大核心
2023年第10期417-421,共5页
Computer Simulation
基金
2022年度教育部产学合作协同育人项目(2206026 90175559)
丽水学院计算机科学与技术国家一流专业(教高厅函[2022]14号)。
