基于多源数据融合的电力作业机载感知设备设计OA

As the requirements for real-time performance and accuracy in power operation scenarios continue to rise,traditional single-source perception methods face significant challenges in complex environments and decision making support.To address this issue,the design of an on-board perception device for power operations based on multi-source data fusion technology was presented.The instrument integrates multi-source data from lidar,Realsense depth cameras,and stereo cameras,and combines multi-source data adaptive rotational convolution(MSDARC)and maximum probability distance intersection over union(MPDIoU)algorithms to significantly enhance target perception and real-time response.The system was applied in a power grid live-line operation robot,where it excelled in tasks such as point cloud stitching and wire recognition,with the maximum time for point cloud stitching not exceeding 0.14 s and an average precision(AP)of 0.852.These results demonstrate the potential of the system for use in complex environments.This study provides theoretical support for intelligent perception in power operations and offers new ideas and methods for applying multi-source data fusion technology in other fields.