基于卷积神经网络的采摘机械臂无碰撞运动规划研究

Research on Collision Free Motion Planning of Picking Manipulator Based on Convolution Neural Network

介绍了卷积神经网络的基本结构及其工作原理,基于DH参数法建立了采摘机械臂运动模型,并设计了一套采摘机械臂无碰撞运动规划算法,旨在实现对采摘机械臂的精确控制。MatLab仿真试验表明:采摘机械臂在系统的驱动控制下,能够准确从起点移动到目标点,轨迹比较圆滑,且能以最优的圆弧路径避开障碍物,优化效果明显,能够满足采摘机器人作业需求,证实了该算法的稳定性和可靠性。

It firstly introduces the basic structure and working principle of convolutional neural network,establishes the motion model of picking manipulator based on DH parameter method,and designs a set of collision free motion planning algorithm of picking manipulator,which can realize the accurate control of picking manipulator.Matlab simulation test shows that the picking manipulator can move accurately from the starting point under the driving control of the system.To the target point,the trajectory is relatively smooth,and it can avoid obstacles with the optimal arc path.The optimization effect is obvious,which can meet the operation requirements of the picking robot,which proves the stability and reliability of the algorithm.