摘要
随着计算机技术的不断成熟和数据分析技术的不断完善,近年来突出机器深度学习功能的智能算法取得重大突破。其中以卷积神经网络为代表的技术,可根据不同的控制要求进行相应数据训练,从而提高系统的控制效果,在机器人控制、目标识别等领域得到广泛应用。随着机器人应用环境的复杂化,设计基于卷积神经网络机器人控制算法在非结构化环境中实现精准化物体抓取,建立一个完整的机器人自动抓取规划系统。
With the development of computer technology and the improvement of data analysis technology,great breakthroughs have been made in recent years in intelligent algorithms that emphasize the deep learning function of machine.Convolutional neural network as a representative technology can carry out corresponding data training according to different control requirements,so as to improve the control effect of the system,which is widely used in robotic control,target recognition and other fields.With the complexity of robot application environment,a convolution neural network-based robot control algorithm is designed to achieve precise object grasping in unstructured environment,and a complete robot automatic grasping planning system is established.
作者
张松林
ZHANG Songlin(Department of Information Engineering,Anhui Institute of Information Technology,Wuhu 241000,China)
出处
《长春大学学报》
2019年第4期14-17,共4页
Journal of Changchun University
基金
安徽省科技厅项目(17030901033)
关键词
机械臂
深度强化学习
策略搜索
卷积神经网络
manipulator
deep reinforcement learning
strategy search
convolutional neural network
