基于优选学习泛化机制的机械臂运动规划方法

Motion planning method of manipulator based on optimum selecting and learning generalization mechanism

针对传统机械臂运动规划方法存在智能化低和适用性差的问题,本文研究了基于优选学习泛化机制的机械臂运动规划方法。根据动态运动基元(DMPS)、机械臂D-H模型和正逆运动学,设计了机械臂示教学习(LFD)运动规划系统。在此基础上依据DMPS的学习特性与样本空间的多样性,提出“方位-距离”筛选规则优选样本,并融入运动特征与障碍物的耦合因子实现避障规划。通过Matlab进行机械臂优选LFD系统建模并仿真分析其可行性与精确性,为了验证系统的适用性,设计并完成障碍物环境下的机械臂避障规划物理实验。本文提出的机械臂运动规划方法在一定程度上赋予了机械臂自主作业的能力,提升了其智能化水平。

Aiming at low intelligence and poor applicability of manipulator’s traditional motion planning method,a motion planning method for manipulator on optimum selecting and learning generalization mechanism is studied.According to dynamic movement primitives(DMPS),D-H model and forward and inverse kinematics of manipulator,a learning from demonstration(LFD)motion planning system for manipulator is designed.On the basis of the learning behaviors of DMPS and spatial diversity of samples,‘direction-distance’filtering rule is introduced to select optimal sample,and the coupling factors of motion characteristics and obstacles is integrated to achieve obstacle avoidance planning.By modeling and simulating in Matlab,the feasibility and accuracy of the manipulator LFD system is analyzed,then the physical experiment for obstacle avoidance planning of manipulator in obstacle environment is designed and completed to test system applicability.In this article,the manipulator motion planning method endows the manipulator with the ability to operate independently,and improves its intelligence level.