preloader
blog-post

The case for Artificial Intelligence in Indoor Farming

author image

Indoor farms promised to provide urban spaces with a bountiful supply of highest quality produce all year round. And yet, for some reasons, the promise remains undelivered.

Once you pry into the field, you will find that the explanation is cost- as of now, the cost of food production due to energy consumption in indoor farms is way too high. More than 25% of the total expense of an indoor farm is electricity cost alone.

For the last two decades, the focus of the industry has been hardware. But, what if, the missing piece is intelligence? More specifically, plant biology and operations intelligence. After all, lab experiments show that it is indeed not only more sustainable and healthier to grow leafy greens indoors, but also cheaper.

Taking total cost into account, our semi-autonomous lab farms are already able to grow high quality produce and yet beat traditional land-based farming cost-wise!

HexaFarms' mission is to make Indoor Farming the norm for sustainable and secure means for urban food production. And that is only possible by outcompeting the market forces.

Our team’s focus is to build the best AI that can take control of the majority of indoor farm operations, from controls to real-time biological activity analysis at a scale limited by nothing but actual farm space. This article outlines what AI brings to the world of indoor farming.

Introduction to CEA and Indoor Farming

From the perspective of plant biology, the sun is just a source of energy in the form of electromagnetic waves, and the earth is just a substrate that holds the plant. Controlled Environment Agriculture is built upon the idea that we can give plants the best growth conditions within a closed environment indoors where plants sit in multiple layers while nutrient is delivered to them via a fluid medium.

Indoor farming is an ideal example of CEA since factors such as temperature, humidity, air composition, light intensity (or even spectrum), nutrient level, etc. can be controlled to an arbitrary precision depending upon the sophistication of the infrastructure. At HexaFarms, for example, we can control all these factors and more.

In CEA, artificial lights serve as the sun (or the necessary ingredient for photosynthesis) and nutrients are readily available in the root zone. A direct benefit of this is that exactly and only what is needed is given to the plants. But, on the other hand, there is (a valid) criticism that indoor farms consume a lot of energy compared to the ‘free’ sun available outside. Even though indoor farms can beat non-CEA agriculture in terms of quality and quanity, they are still struggling with high cost of production though. It is precisely for this reason that indoor farms have relegated themselves to growing niche produce (for now).

A common pattern in an indoor farm showcasing a single-layer of plant-bed

HexaFarms claims to grow food indoors at an efficiency and cost that undercuts even traditional agriculture. We estimate this as a reduction of food production cost by another 30-50% for indoor farms.

The main problem with Indoor Vertical Farming

Energy cost. Indoor farms don’t utilize the ‘free’ sun for a pretty valid and sound reason, but then they also have to pay the cost. The second biggest expense is operations and talent. Because the economy of scale makes sense, most indoor farm investments seem to favour mega-farms (sadly, mega-farms are also the trend for land-based agriculture as well). Of course, AI cannot solve all the problems, but it can address the one that is difficult to solve i.e. plant biology. It is easy to build an LED lamp that can be fine-tuned to a specific spectrum, but it is pretty difficult to know which exact spectrum to use in conjunction with other growth factors for a given plant at a given time.

AI in short

Artificial Intelligence (AI) is a paradigm of enabling a machine to make intelligent decisions. Of course, AI outshines when the problem is too hard for traditional solutions- for instance, cancer detection, driving cars, protein sequencing, drug discovery, generating images from textual description, outcompeting human players at the game of Go, etc.

The scope of problems that AI solves has kept the research community busy since the advent of the field itself. With every passing day, newer and newer challenges are solved by AI in different fields. Our team decided to solve the hard problems of CEA and indoor farming with AI.

HexaFarms is utilizing AI for discovering and maintaining the most ideal growth conditions for indoor food production. The goal of the AI is to minimize the cost of production and to increase growth.

What AI can do for indoor farming

A hi-tech indoor farm is one of the most exciting grounds develop AI applications for. We believe that AI is the last factor that will enable the major reduction in cost of production. Currently, energy requirements and planning require make up for 40% of the production cost[1]. For those starting indoor farming from scratch, if at all, this percentage can very well cross 50%.

HexaFarms' vision to enable just about anyone to become a profitable indoor farmer.

1. Dynamic Plant Phenomics

What if we do not require the grow light to be on for 16 hours? For a mid-sized indoor farming operation, saving an hour’s worth of wasted energy can easily add up to a few thousand Euros in a few months. What if the plants in the grow bed are already stressed and their photosynthetic activity is further worsened by more light? Can a given plant tolerate more extreme temperatures without any compromise on growth? Can a given plant grow more biomass with the same or less amount of resources?. The answer is almost always Yes.

Of the eight factors, let the AI figure out what affects plant health and growth. Note: This is just an example, our usage is more complex and requires other dependencies.

We as species have barely scratched (let alone tamed) the surface of plant biology and various metabolic activities that can occur at the industrial scale, especially when there could be tens of factors affecting plant growth processes. The current process is that meticulous research is done in isolated lab settings, and then the indoor farming operators/managers try to translate the same for thousands of plants in their farms. Not only is this non-interactive and realtime, but also lags in progress that can be made if data from thousands of realtime plants growing in indoor farms can be collected and analyzed.

We use AI for figuring out the exact growth parameters across 14 essential physical and chemical factors that can affect plant growth throughout their lifetime.

A dynamic plant phenomics available at hand would allow us to replicate plant recipes that can then reduce the overall power and resources consumption by exactly and only what a plant needs at a given stage in its lifetime.

2. Talent

Let’s say you want to set up your own vertical farm. You have two options- either you become an expert in indoor farming, or hire a plant expert. An indoor farming ‘expert’ is going to cost you anywhere between 40-80k Euros and will only be able to work for 40 hrs a week. If you have a bigger farm, then you will need more human experts. There are no studies to the best of our knowledge, but it seems that the much sought after and expensive and non-scalable ‘expertise’ can be entirely digitalized.

Our goal is to train and build the best AI system that will surpass any pure human-based expertise to operate hyper-efficient indoor farms.

The main advantage of letting an AI control and operate the farm is not having to rely on human factor and yet getting a guaranteed and ever-improving performance.

In the next article, we will dive deeper into what sorts of things we use AI for and the overall effect on the cost and productivity of indoor farming.

Recent Articles

Let’s stay in touch

We are working hard at the frontiers of so many technologies. If you want to stay updated with our progress, you can sign-up for updates.

*