融合机器视觉和触觉的机器人自动接线方法研究

为保障电力行业的安全生产，降低人工接线操作的危险性和复杂度，提出了融合机器视觉和触觉的机器人自动接线方法。首先，基于改进的Faster R-CNN目标检测网络对导线进行识别，以实现导线端口的精确定位；然后，根据触觉传感器传回的碰撞信息，通过多层感知机（MLP）对碰撞类型进行分类，判断机械臂是否对导线完成了准确抓握；最后，利用三维坐标系对导线端口进行位置修正，并再次通过触觉传感器和带有回归器的MLP指导电线完成最终的接线操作。基于真实开关柜场景进行一系列电力接线实验，实验结果表明融合机器视觉和触觉的机器人自动接线方法可准确完成接线操作，在开关柜中的接线成功率最高可达到80%。所提方法对保障电力安全生产及降低人工接线的危险性具有实际工程意义。

Research on Robot Automatic Wiring Method Integrating Machine Vision and Tactile Sense

To ensure safe production in the power industry, and reduce the danger and complexity of manual wiring operations, this paper proposes a robot automatic wiring method that combines machine vision and tactile sense. First, the wire is identified based on the improved Faster R-CNN target detection network to realize the precise positioning of the wire port. Then, according to the collision information returned by the tactile sensor, the collision type is classified by the multi-layer perceptron (MLP) to determine whether the robot arm has accurately grasped the wire. Finally, use the three-dimensional coordinate system to correct the position of the wire port, and again guide the wire through the touch sensor and the MLP with the regressor to complete the final wiring operation. In this paper, a series of power wiring experiments are conducted based on real switchgear scenes. The experimental results show that the robot automatic wiring method that combines machine vision and tactile sense can accurately complete the wiring operation, and the wiring success rate in the switchgear can reach up to 80%. The proposed method has practical engineering significance for ensuring the safe production of electric power and reducing the danger of manual wiring.