Will agriculture develop a data obesity problem? (Copy)
As the world’s Internet of Things (IoT) sensors now exceed 75 billion devices and data storage is measured in zettabytes, data obesity is becoming a global concern for every industry. In food and agriculture supply chains, the rapid growth in technology, particularly IoT and sensors, risks drowning users in information without insights.
Also known as data bloat, this is the description of massive volumes of data being stored by a business without converting it into clear actions and decisions.
In modern agriculture, tractors, soil sensors, drones, satellite imagery, weather stations, wearable sensors and smart irrigation systems continuously collect data. While data-driven farming promises efficiency, unmanaged data streams create operational and technological friction.
The 5 challenges are:
Paralysis of analysis. From farmers to feed companies, raw data without synthesis overwhelms operators. A grower receiving separate, unintegrated alerts from soil probes, weather platforms and drone flights struggles to determine the priority action for the day. Dairy farmers routinely describe to me how many different alerts cow collars provide, but they focus on just one: heat detection.
Lack of interoperability. Each farm equipment OEM is designing its own proprietary, isolated platforms. They aren’t the only ones, as feed, genetics, management companies and startups get in on the act. The farms and those in the food chain are the ones who suffer. Manually exporting, converting and reconciling spreadsheets means more work, not less.
The limits of connectivity and infrastructure. Rural connectivity is a real issue, and farms in remote areas frequently operate with low internet bandwidth. Uploading gigabytes of high-resolution imagery or point-cloud spatial data to cloud servers can saturate local networks or fail entirely, and a medium-sized livestock farm can generate as much data as a town of people.
The cost of data storage and computing power. Storing terabytes of uncompressed imagery or continuous sub-second sensor feeds has resulted in inflated infrastructure expenses for both agtech startups and agricultural operations.
Reducing the noise, separating the signal. Raw sensor data contains substantial noise (e.g., transient environmental fluctuations or duplicate GPS points). Without edge processing or automated filtering, essential signals (e.g., early disease detection, irrigation leaks) can get buried.
While some agri-investors focus on data as being more valuable than the product, or say that ‘data is the new oil’, I fundamentally disagree. Drowning in data while the fundamental technology doesn’t pay for itself isn’t a solution. So how can we solve this?
Edge computing offers a way forward
Recently, new ag-tech startups have also focused, as they scale, on building systems that compute on the edge, i.e., doing as many computations within the devices as possible, rather than requiring all the data processing to be done off site. An example of this is xSights, the wearable sensor and light for pigs, now in Australia, the US, Europe and Brazil.
As a result of edge computing, the requirements for connectivity and the data sent to the cloud have been reduced through preprocessing. Austria’s Smaxtec is working on the same in dairy farms. The same is true for the explosion in the adoption of feed-bin sensors, where connectivity is clearly a limiting factor in the usability of the devices, e.g., Canada’s Binsentry with over 60,000 sensors installed in North America. The complexity of data has opened the door to SWARM to optimise supply chains and data operability for better overall financial outcomes in commodities, feed, milk and pork.
In summary, to address data obesity, the ag-tech sector is moving away from raw data delivery towards edge computing and decision-support systems. There is no magic pill to ward off this form of obesity; collecting, collating, cleaning and culling data are the keys to avoiding carrying this unnecessary weight, limiting the eventual successful adoption of AI in agribusiness.
https://www.foodagribusiness.world/agribusiness/will-agriculture-develop-a-data-obesity-problem
