Autonomous weed-targeting AI drones a sky-high success
A cost-saving, environment-protecting smart system to target weeds in cotton crops has proved effective in trials at a Central Queensland farm.
CQUniversity’s world-first prototype uses lightweight drones to snap thousands of high-resolution images across a crop.
Then, through local image processing and Artificial Intelligence (AI), weed locations are identified and GPS coordinates are generated, all without the need to upload images to the cloud.
That data is sent to spraying drones using autonomous and localised communication that targets weed locations with herbicide, delivering an innovative solution for green-on-green targeted weed management.
This industry-partnered project, led by CQU’s AI, drone, Internet of Things (IoT), mechatronics and agriculture specialists, held field trials at an Alton Downs cotton farm in February.
The project is led by Dr Jahan Hassan, a specialist in consumer drone services and systems, and Dr Nahina Islam, a specialist in Artificial Intelligence, image processing, and data science.
Following two years of complex development, they confirmed the success of the field trial.
“Our coordinated and targeted spraying system removes the need for blanket spraying of crops – which uses a lot of herbicide and a big investment, and of course has detrimental impacts on the environment,” Dr Hassan said.
“A key innovative approach in our project has been eliminating dependence on the internet for downloading to the cloud – because internet reliance is impossible in many remote locations.”
Associate Professor Biplob Ray led the Internet of Drones (IoD) team to ensure autonomous and internet independence communications.
“Our design means this complex system can work in a localised manner, and that’s a game-changer for how we can support farmers with automating a huge part of their workload,” he explained.
Peter Foxwell is the owner of Foxwell Farms, and has partnered with the CQU team to host the trials.
“My job would be a lot easier if I could sit in the office and control a sprayer, that’s for sure – I’ve got plenty to do on the farm, so this would save time,” he said.
“And if I’m saving a lot of chemicals I’m saving money, (because it) targets weeds specifically – and I’m thinking about the opportunities in other crops as well, why not other broadacre crops, horticulture crops, if there’s value for the farming community I’m all for helping.
“I’m the first farmer to see this stuff happening, partnering with CQUniversity means I get to see amazing technology, and see it first, that’s pretty exciting!”
The success of the trial could shape a cleaner, greener and more cost-effective future for farming.
“The technology needs further fine tuning for commercialisation, but we have a prototype – and we have the results to show the ultimate benefit for farmers, for food security, and for the environment,” Dr Islam said.
The CQU project team also included design and fabrication of mechatronics system led by Dr Abdul Mazid, and weed identification and herbicide efficiency evaluation by Associate Prof Nanjappa Ashwath.
The hard work of Senior Research Fellow Dr Armin Ehrampoosh, Research Assistants Pushpika Hettiarachchi, Dr Anand Koirala, and Dr Mohamed Tolba was instrumental in driving the project to success, alongside contributions from Dr Nurun Nabi, Professor Steven Moore, and Dr Stephen Xu at Charles Darwin University. Dr Zaglul Shahadat from Rajshahi University of Engineering & Technology, Bangladesh, also contributed to this research.
CQUniversity has worked with technology partners OMRON Industrial Automation and Leica Biosystems, and the project is funded by the Commonwealth Government’s Emerging Aviation Technology Partnerships (EATP) program.
