Reducing herbicides to improve our reef

13-row self-propelled sprayer on a Burdekin sugarcane paddock, photo courtesy of AutoWeed's Alex Olsen

Using the latest advancements in artificial intelligence, computer vision and robotics, Mostafa Rahimi Azghadi and his team, in collaboration with industry partner, AutoWeed, Sugar Research Australia, and Department of Agriculture and Fisheries Queensland, are aiming to improve water quality in reef catchments.

This innovative project has demonstrated significant water quality improvement. The AutoWeed smart spot spraying technology reduces the concentration of pesticide found in runoff, compared to the current blanket spray approach, while matching the same level of weed knockdown.

Smart spot spraying vs Blanket spraying

The project has completed 6 successful spray trials and has shown that on average robotic spot spraying on sugarcane farms is 97% as effective at weed knockdown while reducing herbicide usage by an average of 35% across all trials, compared to the industry practice of blanket spraying. In the best case, where weed pressure was lower, spot spraying was found to reduce herbicide usage by up to 75%. This includes specific trials on nutgrass in ratoon cane, nutgrass in plant cane, grass weeds in a beans crop and broadleaf weeds in a beans crop. Furthermore, strip replicated end-of-row water quality assessments showed that spot spraying reduced the mean concentration and mean loads of herbicides detected in runoff by an average of 61% and 46%, respectively, compared to blanket spraying. These results also confirm that herbicide concentrations and loads in runoff are directly proportional to the amount of herbicide sprayed. This reveals the significant potential of smart spot spraying to provide a step change in reef water quality if used at scale.

Mostafa Rahimi Azghadi and Alex Olsen, presenting their project results to the Mulgrave Russel growers Water Quality forum in Mirriwini

Auto-weed spray trial team in the Burdekin

A long-term objective of the project is to facilitate wide-scale adoption of the AutoWeed technology for sugarcane weed management. The existing proof-of-concept project has demonstrated that wide-scale uptake can result in a step decrease in the herbicide usage across the sugarcane industry, leading to improved water quality for the environment and the Great Barrier Reef (GBR).

JCU Auto-weed machine

To achieve this long-term objective, Mostafa and the team are currently

(i) extending the spray trial work to more weed species across the Burdekin region and dataset collection to the Mackay/Whitsundays and Wet Tropics regions,

(ii) improving the AutoWeed technology with valuable new features (such as lighting, dual herbicide spraying and selective pre-emergent application) based on the requirements of the growers, and

(iii) providing a clear economic analysis of the use of the AutoWeed technology, showing the growers the financial and environmental benefits they can achieve in the long-term, should they adopt the AutoWeed technology.

This $900K collaborative project between James Cook University, AutoWeed Pty Ltd, Sugar Research Australia, and Queensland Department of Agriculture and Fisheries is operating between 2020 and 2024.

This project is funded by the partnership between the Australian Government’s Reef Trust and the Great Barrier Reef Foundation.

Contact details

Project Research Publications

• Low-Power and High-Speed Deep FPGA Inference Engines for Weed Classification at the Edge
• DeepWeeds: A Multiclass Weed Species Image Dataset for Deep Learning
• Robotic Spot Spraying of Harrisia Cactus (Harrisia martinii) in Grazing Pastures of the Australian Rangelands