Machine Learning research project begins at NM Group

Posted

Last month, NM Group has begun a new Knowledge Transfer Partnership (KTP) with Durham University. The new research will look at the use of Machine Learning and Deep Neural Networks to innovate feature extraction from LiDAR data. This is the third major research project with the university following two previous KTP initiatives.

Since June, our KTP associate Nan Wang has been looking at ways of applying Machine Learning to aerial LiDAR data of powerlines. Although automatic extraction techniques exist, none are sophisticated enough to achieve the accuracy needed for the complex powerline environment. This project aims to use expertise and new research from Durham University to advance the technology and bring it into industry.

Before undertaking new work on applying machine learning, Nan will begin by assessing existing studies on the topic. This includes studies on advanced neural networks. This includes research from institutions such as Stanford and industry papers from Apple Inc. Some of the challenges are that most previous work has not used aerial LiDAR data. Additionally, due to the cutting-edge nature of the technology, few commercial tools exist.

Commented on the KTP, Nan added, “I am really excited to join this project and looking forwarding to applying academic research advancements to industrial productions. Deep learning solutions for processing 3D point clouds have been developed in academia. However, none of them have been tested on large-scale aerial LiDAR data as seen at NM Group. This research project will be beneficial for both NM Group in terms of efficiency and Durham University in terms of validating academic research achievements in a real-world setting”.

There is much work to do, but when the outcomes could lead to significantly faster data analysis for power utilities (and therefore safer, more reliable networks) the effort is clearly justified.

Read more about NM Group’s other KTP in vegetation management for power networks.