Strategic Elements Ltd (ASX:SOR) subsidiary Stealth Technologies has successfully demonstrated the potential of its automation and robotics platform (AxV) to be leveraged for the multi-billion dollar global agriculture sector.
Work with collaborators the Australian Herbicide Resistance Initiative and the University of Western Australia School of Agriculture and Environment has assisted the company to achieve early stage validation. Further optimisation and engineering leading to an expanded technology demonstration across multiple potential end user reference sites is to be conducted.
Managing Director, Charles Murphy, said Stealth has also filed patent protectioncovering a weed detection device, including the arrangement of sensors on a moving vehicle platform, algorithms, methods of weed detection, and software to process crop data from a moving vehicle to estimate locations of weeds.
Available advanced weed detection technologies typically use RGB cameras and different forms of imaging that distinguish weeds and crops via colour. This has serious limitations in broadacre cropping where weeds are often the same colour as crops.
The company is taking a different approach by leveraging the sophisticated sensor, mapping and localisation technology already built and used in its Autonomous Security Vehicle collaboration2 with US Fortune 100 Company ‘Honeywell’.
Mr Murphy said the need for excessive use of chemicals and production loss costs are significant issues for the global agricultural industry. The estimated cost of weeds in Australian cropping systems alone is at A$3.3 billion annually. Total annual cost of weeds in the United States are estimated at US$34.5 billion.
Stealth, UWA and AHRI collected logistics and in-field scoping data from a large scale broadacre farm in WA to enable detection of weeds protruding above the canopy of a barley crop. Weed detection prototype hardware was developed and installed onto a combine harvester during harvest.
Algorithms were then developed, tested and validated by comparing the location of weeds detected by the technology with known locations of weeds with visual confirmation. On a limited data set the technology was able to detect 100% of weeds with a height threshold of 20cm above crop canopy.
Significantly, the technology was able to detect weeds from the barley crop notwithstanding the fact that both were brown in colour and barely distinguishable to the human eye. Currently available technologies using computer vision solutions would be unable to replicate these results.
Traditionally, farmers have treated weeds in large-scale crop farming with the mass universal application of herbicides, however weeds are becoming herbicide resistant. This means that farmers have to use targeted herbicides that are more expensive and can lead to new generations of weeds resistant to these new herbicides. A solution is to use alternative agronomic techniques to target weeds. This can include spot application of herbicides, increasing the seed rate over known weed areas when planting crops to starve out weeds during the season and increasing fertiliser rates within paddocks over weed locations during the season.
All these techniques require accurate knowledge of weed locations, that to date is lacking in farm management. The technologies that Stealth and its collaboration partners are developing are designed to solve this problem in large-scale crop farming whilst keeping the work-rate at the required levels.
The value proposition for Stealth is to deploy this technology to farms around the world where large-scale crop farming exists. This includes not just crops such as wheat and barley but could be extended to corn, canola, and other large-scale crop types. The usage of this technology together with agronomic techniques that can leverage weed location knowledge could dramatically decrease herbicide input costs to farming whilst maximising crop yields making farming more efficient and profitable for farmers.
Milestones and Schedule of Work for the Collaboration
- Optimisation of the weed detection technology – further work to optimise the technology for different environments, crop types and deployment scenarios. Q2
- Product engineering – advancing the technology into a versatile and easy to use package that can be installed by farmers onto a wider range of farm equipment – from combine harvesters to sprayers and other farm equipment. Q3 and Q4
- Demonstration scale-up – increase end user reference sites at multiple sites and in different environments. A program is being developed to deploy the weed detection technology to at least ten potential end customer reference sites in the November 2021 harvest. Q3 and Q4
Our strategy to build a platform that had applications across multiple industry sectors is starting to fulfil its promise. Our commercialisation strategy is to collaborate closely with end users to solve a real, existing problem with automation,” Mr Murphy said.
“From an Australian domestic market context other sectors like logistics and mining also have attractive opportunities and we are very active in seeking the right partners with which to collaborate.”
