摘要
为了合理补偿机器人定位误差,提升作业能力,该文提出基于深度学习网络的机器人定位误差估计与补偿方法。确定机器人定位采样点,获取机器人末端定位理论位姿,以机器人末端理论位姿作为深度神经网络输入量,机器人末端定位误差作为输出量,利用遗传粒子群算法优化权值与阈值,得到机器人定位误差估计值,并对理论位姿坐标反向迭加该误差估计值,完成定位误差补偿。实验证明,该方法能够有效补偿机器人的位移偏差和关节角度偏差,精准抓取目标物体,并在不同数量采样点条件下,可使不同类型的机器人保持较高的定位精度。
In order to reasonably compensate for robot positioning errors and improve operational capabilities,a deep learning network-based robot positioning error estimation and compensation method is proposed.Determine the sampling points for robot positioning,obtain the theoretical pose of the robot’s end positioning,use the theoretical pose of the robot’s end as the input of the deep neural network,and use the robot’s end positioning error as the output.Use genetic particle swarm optimization algorithm to optimize the weights and thresholds to obtain the estimated value of the robot’s positioning error,and reverse stack the estimated value of the theoretical pose coordinates to complete positioning error compensation.Experiments have shown that this method can effectively compensate for the displacement deviation and joint angle deviation of robots,accurately grasp the target object,and maintain high positioning accuracy for different types of robots under different number of sampling points.
作者
田立国
熊磊
TIAN Liguo;XIONG Lei(School of Automotive and Mechanical and Electrical Engineering,Hanzhong Vocational and Technical College,Hanzhong 723000,China)
出处
《自动化与仪表》
2023年第7期38-41,46,共5页
Automation & Instrumentation
关键词
深度学习网络
误差估计
误差补偿
定位采样
深度神经网络
遗传粒子群算法
deep learning network
error estimation
error compensation
location sampling
deep neural network
genetic particle swarm optimization
