基于神经网络的遥操作工程机器人力觉反馈技术

Force-feedback Technique of Tele-operated Construction Robot Based on Neural Network

提出基于神经网络模型的摇操作主从式工程机器人的力觉反馈技术原理。首先根据机器人从动手机构的空载动力学特性,离线建立了空载神经网络模型,然后依此模型在线辨识空载驱动力以间接提取负载力。该方法可提高力觉反馈的准确性和同步精度,实验结果验证了其有效性。

The force - feedback technique principle based on neural network model for master - slave system of Tele - operated construction robot was proposed. First, the no - load dynamics neural network model was built off - line based on the no - load dynamics characteristic of the slave actuator of the robot. Then, the no - load dynamics neural network model of the slave actuator was used on - line to identify no - load driving force so as to obtain the load force. The accuracy and synchronicity of force - feedback were obviously improved. The experimental results confirm the usefulness of proposed method.