Project 6.07 Airborne Powerline Inspection Technology Improvements


Precision Guidance of Aircraft

Ergon Energy has over 690,000 customers spread over 1.7 million square kilometres, or 97% of the state of Queensland. Their principal asset is over 150,000km of powerline and almost 1,000,000 power poles. A significant component of their cost of supply is in the condition monitoring of the powerline assets and the vegetation in the powerline corridor. Each year Ergon spends $80 million on inspection and clearing vegetation. The three year CRC-SI project 6.07 for Spatial information business improvement applications at Ergon Energy focussed on precision guidance of aircraft over powerlines, asset and vegetation detection using LiDAR imagery, and detection and classification of tree species. The project was successfully completed in 2010 and provided the fundamental basis for the FAS project and the 4.3.1 eFAS project.

Milestones & Achievements

Powerline 3D

Powerline and Vegetation Detection in LiDAR and Imagery

  • Route Optimisation Techniques: The ARCAA team developed new route optimisation algorithms for powerline inspection aircraft based upon biologically-inspired meta-heuristic techniques were further developed and refined (calculated inspection routes involving 20% shorter flight time than human planned routes). These algorithms were characterised against alternative planning approaches.
  • Precision Guidance of Aircraft: Cost-effective airborne vegetation surveillance requires precision guidance of the survey aircraft in order to keep the sensors scanning the target area. New guidance algorithms for the related problem of inspecting a sequence of isolated point assets (such as power pole isolators) were developed and evaluated through simulation and flight tests.
  • Powerline and Vegetation Detection in LiDAR and Imagery: The aim of the automated processing is to automatically and robustly extract vegetation management clearances, thus saving on the labour costs and effort associated with manual processing of voluminous data sets. New LiDAR automatic power line detection techniques were developed, and were shown to have superior detection rates compared to commercially available LiDAR processing products (independently tested). A number of state-of-the-art local feature descriptors and machine learning classifiers were also evaluated in order to select the most appropriate features in tree species classification; a new rotation and scale invariant spectral-texture feature descriptor has also been developed and evaluated.

Project Contacts

For further information regarding this project please contact the project contact below:

  • Professor Duncan Campbell
  • Dr Jason Ford