Improving the accuracy of weed killing robots with new image processing algorithms and near infra-red spectroscopy techniques (Old ID 23146)
Role
Chief Investigator
Description
Automated weed species recognition remains a major obstacle to the development and industry acceptance of robotic weed control technology. Particular problems occur in rangeland applications, including high light variability and weed-camera distance variability, which cause camera dynamic range problems, image blurring, and occlusion by other plants. This project aims to develop robust image recognition systems combined with Near Infra-Red spectroscopic methods for these complex rangeland environments with special emphasis on the broad-acre grazing pastures in North Queensland. The developed imaging systems will be suitable for all weed killing applications with particular emphasis given to foliar spot-spraying and Herbicide Ballistic Technologies.
Date
26 Jun 2017 - 18 Oct 2019
Project Type
GRANT
Keywords
Weed;Near-Infra-Red;Robotics;Image Analysis
Funding Body
Australian Government Department of Agriculture, Water and the Environment
Amount
271772
Project Team
Alex Olsen;Owen Kenny;Bronson Philippa