Research_

Precision Weed Control Group

Site-specific weed control solutions

Utilising innovations in technology, the research group is developing site-specific weed control capacity for Australian agriculture systems. With expertise in weed biology, agronomy and machine learning.

Our aims

The overall objective for the Precision Weed Science Group is the development of alternate weed control technologies based on site-specific approaches. The specific research aims are:

Development of site-specific approaches to weed control in cropping systems
Identification of weed recognition techniques
Evaluation of weed identification and detection technology
Evaluation and development of alternate weed control technologies
Determine the efficacy of in-crop and strategic weed control technologies.

Our research

The Precision Weed Control group’s research efforts focus on innovation in weed control towards the introduction of alternative weed control systems for Australian agriculture. Utilising new sensing technologies coupled with software developments in machine learning, the group have initiated weed recognition research for Australian weed species.

Weed recognition enables site-specific weed control and parallel research is exploring the efficacies of alternative precision control treatments such as lasers and electrical weeding. One of the few in-crop weed control techniques that can routinely be used in cropping systems is cultural weed control.

Our research efforts are aimed at enhancing the crop competition effects on weed growth and reproductive development through the use of narrower row spacings and higher plant densities. As well as the development of in-crop weed control technologies the use of disruptive weed control solutions are being evaluated for strategic use as an intervention to escalating weed population densities. Towards this the impact of cover crops as a disruptive weed control strategy are being evaluated.

Current projects

Project Title: Innovation in Crop Weed Control
This five-year (2016-21) GRDC funded project is focussed on the development of weed management techniques and strategies that allow the mitigation and avoidance of problems associated with the widespread occurrence of herbicide resistance in the northern grains region of Australia. This research, lead by University of Sydney, is being conducted in five key areas involving five research agencies, University of Sydney, New South Wales-Department of Primary Industries, University of Queensland, Department of Agriculture and Fisheries Queensland and Charles Sturt University:

i) Herbicide Innovation – Assoc. Prof. Michael Walsh, University of Sydney
ii) Crop Competition – Dr. Michael Widderick, Department of Agriculture and Fisheries, Queensland
iii) Strategic Weed Control - Prof. Leslie Weston, Charles Sturt University
iv) Investigative Weed Biology – Assoc. Prof. Bhagirath Chauhan, University of Queensland
v) Engineering Weed Control Solutions – Assoc. Prof. Michael Walsh, University of Sydney

Project Title: Site-specific Weed Control for Ginger Cropping Systems
Ginger production systems require labour intensive manual weeding during the critical weed free period from August to January, a costly approach for weed management. This AgriFutures Australia funded project is seeking to improve weed control opportunities in ginger production systems that significantly reduce costs and application risks associated with herbicide-based, manual weed control. Towards alleviating these risks the specific aims are i) development of effective weed identification algorithms suitable for real-time use, ii) identification of alternative herbicide treatments for site-specific application and iii) development of site-specific delivery technologies suited to autonomous use in ginger crops.

Project Title: Identifying a machine learning approach to weed recognition in Australian grain production systems
The revolution in artificial intelligence (AI), made possible by access to large datasets and improved computing capabilities, has enabled the opportunity for weed recognition and subsequently site-specific weed control for the Australian grains industry. Large annotated image datasets and suitable weed recognition algorithms are required for in-crop identification of Australian weeds. With weed recognition enabling technologies only recently developed there is a need to define suitable approaches for both data base and recognition algorithm development. The objectives of this project are to i) initially evaluate ML approaches for weed recognition, then ii) identify a suitable framework for an open source image library, iii) implement the open source image library and iv) use the identified ML approach to development weed recognition algorithms for key weed species.

Project Title: Managing awnless barnyard grass by crop competition

Barnyard grass, Echinochloa spp. (BYG) is responsible for significant yield losses in Australian cropping systems. Reliance on glyphosate for barnyard grass management in fallow and cotton cropping systems has resulted in the widespread evolution of glyphosate-resistant populations across the northern cropping region. Non-chemical control strategies are needed for more effective herbicide barnyard grass management. This masters/Ph.D. project aims to better understand the role of altered agronomic practices in enhancing the mung bean competitiveness against barnyard grass. More specifically, the effect mungbean planting time and planting density on the growth and seed production of awnless barnyard grass will be investigated under field conditions.