Better decisions

The ability to use software to make or support decisions, even in the absence of complete information, will help mitigate risk for many operations. Artificial intelligence can process the vast amounts of data generated by farm machinery and combine it with historic data to make management recommendations. In the first instance, the farmer may augment these recommendations with their own knowledge and experience to refine decisions. However, as machine learning records the characteristics of each paddock over time, it will adapt its recommendations, which will progressively improve decision-making beyond human capability

Reduced operating costs

In agriculture, every paddock is different every year, with a range of variables impacting productivity and profitability. The traditional agricultural practice was to treat every paddock in the same way, with universal applications of inputs, like chemicals and water. Over the last decade or so, variable rate application and irrigation scheduling have enabled tailoring of input application to even out crop variability, and lift yields and efficiency. Artificial intelligence will enable management and treatment decisions to be tailored to individual plants and animals, further reducing input costs and lifting overall productivity. Artificial intelligence can be used to undertake rapid nutritional needs analyses of individual animals based on behaviour, body condition score and feeding rates. Understanding optimal feed rates and needs will reduce waste. Robots using artificial intelligence can identify weeds from seedlings and apply herbicide directly to the weed. This reduces the amount of chemical applied across the paddock and reduces the economic and environmental impacts of spray drift. Australia’s high cost of labour impacts agriculture’s global competitiveness. This is particularly true in horticulture, where 50–70% of production costs are spent on labour intensive tasks like weeding and picking. Commercial installations of artificial intelligence to farm machinery could automate these roles within 10 years. This will change the business model within horticulture from the existing emphasis on operating costs to capital costs as additional machinery is purchased. As at 2016, it is estimated these capital outlays would deliver a two-year return on investment.

Changed skills needs

Many agricultural industries claim productivity losses are, in part, caused by labour costs and shortages. Artificial intelligence may supplement businesses that struggle to find a reliable and skilled workforce. There is agreement that artificial intelligence will create disruption within employment markets in a range of industries, including agriculture. This creates concerns about the displacement of the agricultural workforce (particularly low and middle-skilled workers) and the negative impact on future prospects for careers in agriculture. Artificial intelligence will drive changes in the skills required by the agricultural industry, and there will be a stronger emphasis on analytical and information technology roles, rather than manual labour or machine operation. A 2015 report Future workforce trends in NSW: Emerging technologies and their potential impact considered how computerisation and automation will affect jobs in New South Wales over the next 10–15 years. The probability of job losses for agricultural and forestry scientists was only 5.4%, compared with agricultural technicians at 56.8% and a 91.2% probability that agricultural, forestry and horticultural plant operators will be replaced by automation in the same period. However, it is expected that technological changes, which are already underway in many industries, will occur incrementally allowing the existing workforce to diversify and refine its skills concurrently with the change in technology. The adoption of artificial intelligence will also attract a new generation to the agricultural workforce to build careers in the sector, with skills and experience beyond agriculture.

Managing individual plants and animals

Artificial intelligence is critical for crop management robots to discriminate between crop seedlings and weeds. The University of Sydney’s prototype robot, RIPPA (Robot for Intelligent Perception and Precision Application), navigates through horticulture crops using a global positioning system (GPS). Sensors scan the ground as it moves, collecting data and passing it through machine-learning algorithms to classify each plant as a weed or seedling. Using VIIPA™ (Variable Injection Intelligent Precision Applicator) technology, RIPPA can autonomously apply herbicide directly to the weeds only. Researchers at Queensland University of Technology combined UAVs with remote sensing and machine learning to assess a sorghum crop severely damaged by white grubs (family Scarabaeidae). Currently, crop damage is assessed by farmers or agronomists who categorise it based on visual observation. A UAV fitted with imaging sensors and using machine learning was able to assess the damage in accordance with existing classifications, but it simplified the process by minimising manual input and human error. In time, researchers believe UAVs could assess crop damage faster and more accurately than humans.

Researchers at the University of Technology Sydney have used artificial intelligence to analyse 3D images of muscle and fat on cattle to determine an accurate condition score. The images are taken by sensors as cattle move through a crush, providing farmers with real-time information about each animal’s condition and suitability for the market.

Increasing productivity with patterns

Artificial intelligence software can identify biophysical patterns that cannot be seen or captured by humans. The University of Sydney’s Mantis and Shrimp robots were able to identify the best locations for the planting of pollination trees in orchards by analysing the biophysical data to determine low yielding areas within the rows. Researchers from the US, Iran and Australia have used machine learning to rapidly filter datasets seeking patterns for the physiological and agronomic factors that will increase maize yield at a regional level and at the paddock level.

Understanding our world

In Europe, the PLANTOID project is developing artificial intelligence based on the sensing and adaptive behaviour of plant roots in response to their environment, particularly the way they seek nutrients and moisture with minimal use of energy. The goal is to develop soil exploration and monitoring robots, which in future may be used to detect heavy metals in contaminated environments or even determine the composition of the surface of other planets. Being able to mimic a plant’s sensory capabilities could also be adapted for use in smart devices that can operate autonomously. The Green Brain project in the UK is creating a computer model of the systems in a honeybee’s brain that governs vision and sense of smell. The ultimate goal is to deploy a robot that thinks, senses and acts like a honeybee without human intervention. Researchers have successfully uploaded the software to a UAV, which flies autonomously using the vision software. This could result in the development of robot bees to autonomously pollinate agricultural crops. Being able to model the complete brain of an organism will also pave the way for future research into understanding animal behaviour, which could support a range of artificial intelligence applications in agriculture. One example is computer simulations to understand optimum stocking rates or housing design in intensive agriculture systems.

Automated irrigation systems

As any plant grower knows, traditional irrigation management is an arduous task. This is coupled with a heavy reliance on historical weather conditions to predict required resources. Thankfully, though, automated irrigation systems are designed to utilise real-time machine learning to constantly maintain desired soil conditions to increase average yields. Not only does this require significantly less labour and have the potential to drive down production costs, but with 70% of the world’s freshwater used for agriculture, the ability to better manage how it’s used will also have a huge impact on the world’s water supply.

Crop health monitoring

Similarly, conventional crop health monitoring methods are incredibly time-consuming and are generally categorical in nature. In comparison, companies developing automated detection and analysis technologies – such as hyperspectral imaging and 3D laser scanning – will substantially increase the precision and volume of data collected. With the ability for microscopic data collection, farmers will be able to produce diagnostics specific to individual plots or even single plants.

Facial recognition

Facial recognition is nothing new, however, this is now extending beyond humans into the world of domestic cattle. Whilst ‘smart’ cattle monitoring is more commonplace, existing systems largely require the use of physical tracking devices. Facial recognition technology will eliminate the stress of fitting these devices, allowing easy monitoring of an entire heard with minimal interaction. This is set to enable individual monitoring of group behaviour, early detection of lameness and accurate recording of feeding habits.

Although hailed as the future of farming, the extent to which AI will change the daily operations of the traditional family farm is yet to be seen. However, with new Agritech companies producing increasingly accessible technology, the ‘digital farm’ of the future may be closer than we think.