Research and Analysis of Robot Dynamic Visual SLAM Based on Live Working Scene of Distribution Network

Abstract

Due to the rapid development of artificial intelligence technology, the demand for smart distribution network powered robots is becoming stronger. In the live operation scene of the robot distribution network, due to the influence of external natural conditions, the wire is not completely static, but the accuracy of the wire identification directly affects whether the operation can be performed normally. Dynamic objects such as manipulator arms of the distribution network live working robot will interfere with the positioning and mapping of the robot itself during the operation, which will affect the normal operation. In response to the above problems, this paper selects and analyzes the semantic segmentation method based on deep learning to identify dynamic objects and wires, obtains the wire depth estimation from the camera model and the semantic segmentation results, and finally uses the visual SLAM algorithm to complete synchronous positioning and mapping.