Prof Mostafa Rahimi Azghadi

Title: Improving water quality in reef catchments through AI-enabled precise robotic weed control on sugarcane farms

Abstract:

Widespread herbicide use in sugarcane farms poses a major risk to water quality in the Great Barrier Reef (GBR) catchments. To address this, we developed AutoWeed, a robotic spot-spraying system powered by deep learning for weed detection and precision application. We trained AI models to identify priority weeds such as nutgrass and Guinea grass. The system was retrofitted to existing farm machinery and trialled across 86 hectares over four years in key Queensland sugarcane areas. Field results showed AutoWeed achieved 96% weed control efficacy compared to 99% for traditional broadcast spraying, while reducing herbicide use by 44% on average and up to 78% in low-weed-pressure areas. For hard-to-control Guinea grass, AutoWeed proved especially effective, offering a valuable alternative to manual spot spraying. Importantly, runoff monitoring revealed a 46% reduction in mean herbicide concentrations compared to broadcast spraying, demonstrating AutoWeed’s potential to enhance water quality outcomes in GBR catchments through smarter, targeted weed control.