论文部分内容阅读
针对机械臂行为动作获取的问题,设计了具有模仿学习机制的机械臂系统,并在ADAMS下搭建了其等比例运动学模型。通过拖动仿真模拟手把手示教过程,将采集到的示教行为信息用以表达模仿学习策略的BP神经网络进行训练,获得示教行为信息间的映射关系,应用于机械臂系统实现对示教行为动作的快速学习。结果显示:该方法具有良好的行为编码能力,能够实现机械臂连续动作的模仿,故本系统对机械臂行为动作的获取是可行的,解决了目前机械臂系统动作规划编程复杂的问题。
In order to solve the problem of manipulative behavior of manipulator, a manipulator system with imitation learning mechanism is designed and its proportional kinematics model is built under ADAMS. By dragging the simulating hand-handle teaching process, the collected teaching behavior information is used to express the imitation learning strategy BP neural network for training, and the mapping relationship between the teaching behavior information is obtained, which is applied to the robot arm system to realize teaching Quick learning of action. The results show that this method has a good behavior coding ability and can imitate the continuous motion of the manipulator. Therefore, this system is feasible to acquire the manipulative behavior of the manipulator, and solves the problem of complicated programming of the manipulator system.