Not only can artificial intelligence reduce environmental impact in agriculture, combat weeds and optimise production, AI may also be what is needed to prevent food shortages due to climate change and potential trade crises.
Rising global temperatures can create problems for agriculture across Europe, including Sweden; more severe drought in some regions, higher precipitation in others, as well as fluctuating temperatures, hours of sunshine, winds and humidity. Local ecosystems, with their microorganisms, microbiota, and pest insects, may change – along with the soil’s ability to produce crops.
– “There may be large fluctuations in harvests in the future. An important role for AI will be to make the production chain in agriculture more robust,” says Fredrik Viksten.
He is a former researcher at RISE and an expert in image analysis and control of robots with camera systems. Image analysis from agricultural fields is key to taking control of crop well-being and growth.
– “This spring we will be running a project in our testbed outside Uppsala where we will use AI to determine the precise growth stage of a crop in real time, and based on that we can implement the right measures at the right time.”
The project exemplifies how data from agricultural fields, such as photos taken by drone, can be combined with machine learning AI to improve the odds of a good harvest.
– “Image analysis is not entirely straightforward. For example, before the AI learns what stage of growth a camera is showing in real time, it must learn what a crop looks like at each stage. There are approximately one hundred stages of growth, so access to many images is required. We have access to about a thousand previously taken images at each stage at different light conditions and angles, and these must be tagged manually by experts in crop development. This is necessary before the AI has learned to tell the difference between the different stages.”
Camera-based AI technology, along with autonomous robots, can also be used for weed control. One vision is that small combine harvesters or weeding robots could operate in the fields around the clock, like robotic lawnmowers at home in the garden. This makes it possible to have small-scale fields with large-scale efficiency. Narrowing fields and interspersing them with different types of crops maintains quality and productivity, and strengthens biodiversity.
The more data, the better
Access to data is crucial
In the development of AI systems that can help agriculture in new climate conditions, access to data is crucial.
– “Traditionally, a farmer has been able to do quite well learning from personal experience from previous harvests. But as the climate changes, past experience no longer matches the new growing conditions. In this, AI can make a big difference, since it’s theoretically possible to collect metrics and experiences from 10,000 farms from different years, and then draw conclusions from that data regarding necessary measures, including which crops should be grown, in order for farmers to have bountiful harvests in new conditions.”
Viksten envisions how farmers will be assisted by a sophisticated AI control panel that helps to make the right decisions for a successful harvest.
More investment is needed
According to Viksten, more investment will be needed to develop and scale up the AI systems that farmers need:
– “The important thing is to get data collection up and running, since you can only get data from growing seasons and harvesting 1-2 times a year. For the AI to draw the right conclusions using data in ten or twenty years, sensors must be out into the fields now, as soon as possible. The least we can do for the future is to start recording temperatures, hours of sunshine, and rain on all farms in Sweden, preferably all over Europe – the more data, the better.”
CENTRUM FÖR TILLÄMPAD AI PÅ RISE
Centrum för tillämpad AI på RISE bedriver avancerad forskning inom AI, kopplar samman expertis och applikationer inom RISE och utforskar ett brett spektrum av innovativa applikationer med företag och offentlig sektor.