The Future of Smart Greenhouses Empowered by AIoT

Sophia Yu
Foothill Ventures
Published in
9 min readJul 15, 2022

– Why Foothill invested in Koidra and how it will change modern greenhouse operations

We see a lot of “AI-enabled industrial/ agricultural automation” companies. In general, these take the form of (a) integration with sensors and controls (often employing computer vision); and (b) superior data analysis to improve decision making. The benefits are multi-faceted, including improved yield, lower labor costs, preventative maintenance, and lower cost inputs.

Even though all of the pieces of the story are attractive to us, we almost never pull the trigger on these deals. Ultimately, we end up concluding that there is no barrier to entry in these businesses until they get to scale (and have access to more data than any other player). In this respect, there is a chicken-and-egg problem: they do not have more data until they get to scale, but they may not make it to scale because of a lack of data advantage.

When we first saw Koidra, we had a similar reaction: an extremely smart team (led by Kenneth Tran, an MSFT AI research veteran) that had achieved stunning results in the Autonomous Greenhouse Challenge in 2018, but no compelling reason for sustained advantage. However, upon digging into the company’s story further, the following seemed obvious to us:

  1. “Controlled Environment Agriculture” (“CEA”) — including greenhouses — will be more greatly impacted by automation than conventional agriculture
  2. CEA will continue to gain share against conventional agriculture, as their cost and productivity curves are far steeper than those of conventional agriculture
  3. The CEA market is relatively underdeveloped in the US relative to Europe (and also vs. Canada and Mexico); this will change as labor pressure increases and technology improves
  4. The agricultural market is highly technical, with growers paying close attention to any breakthrough that can improve efficiency and yield
  5. The wins achieved by the Koidra team at the Autonomous Greenhouse Challenge were well-known by all major growers. The amount of inbound interest in working with Koidra is so overwhelming that they can quickly solve the chicken-and-egg problem.

To understand why we have confidence in Koidra, it is useful to first understand what they have achieved at the Autonomous Greenhouse Challenge

First, congratulations to “Team Koala” for winning the Autonomous Greenhouse Challenge a second time!

Automation is a clear trend across many different industries, and it is especially true for greenhouses as you can essentially control all the parameters in the controlled environment. Autonomous greenhouse technologies can lead to a larger production of fresh and healthy vegetables, improved food safety, and fewer resources used in the production.

The Koidra-led team Koala, consisting of the startup Koidra, Cornell University researchers, and the Dutch company Van der Valk Hortisystems, won the 3rd edition of the Autonomous Greenhouse Challenge at Wageningen University & Research (WUR) in Bleiswijk. Kenneth Tran, who is the team captain and also founder and CEO of Koidra, also led the winning team of the 1st edition of the Autonomous Greenhouse Challenge in 2018. In fact, Kenneth founded Koidra after winning the 1st edition of the Challenge.

In this 3rd edition, five international teams, selected from 46 teams from 24 countries, created their own fully autonomous algorithms to produce lettuce. The goal is to produce good quality lettuce and minimize possible inputs and resources such as energy and CO2, thus optimizing the net profits.

With fully autonomous algorithms, greenhouse sensors, and 3D cameras, the system will automatically determine the environmental elements such as temperature, light (including daylight and artificial light), heat, and CO2. It also sets parameters for cultivation-related decisions such as density, spacing, and day of harvest.

Team Koala’s results are superior. It is worth noting that Team Koala is the only AI team that outperformed the reference grower who came second, while the rest of the AI team could not beat the human growers. The net profit of the team is 5.93/m2 vs. 4.64/m2 for the grower team. It also achieved the lowest cost, 7.24 vs. 7.26 from Veggie Might whose total cost is the second lowest vs. 7.49 from the grower.

Results from the 3rd AGC. Resource use per kg sellable lettuces by different teams
Results from the 3rd AGC. Team Koala achieved the highest net profits

Current greenhouse market

Greenhouses are an important part of global agriculture, accounting for over $350B of crops in 2017. The US, Canada and Mexico together have over 43k acres of greenhouse farms. It is worth noting that greenhouses in North America are underpenetrated relative to Europe, and that the US is underpenetrated relative to Canada and Mexico. We believe that greenhouses will continue to increase their share of global agriculture because of the degree of control that they can exert over their environment — simply put, their costs and yield improvement curves are steep enough that they will almost inevitably become more profitable in many crop categories in many parts of the world. In the US, several large tech-enabled greenhouse farms have been founded in recent years, such as AppHarvest and Local Bounti.

Greenhouse growers have been looking for solutions to improve production efficiency, increase yields and automate processes. Agriculture as a whole is one of the most data-driven businesses in the world, and growers are often very early adopters of automation technologies. The increase in productivity on the farm in the past hundred years has been stunning, and well beyond many other industrial sectors.

On the other hand, industrial automation systems are still fairly rudimentary. The industry is dominated by a small number of Dutch industrial control systems that are primarily hardware companies. Based on our customer reference calls and market research, the current market leaders are:

  • Priva: based in the Netherlands and founded in 1959, selling equipment and software for climate and process control in horticulture
  • Hoogendorn: based in the Netherlands and founded in 1974, focusing on digital horticultural automation systems
  • Ridder: based in the Netherlands and founded in 1974, technical control solutions for protected horticulture
  • Argus Controls: based in Canada and founded in 1984, building environmental control and automation solution for horticultural operations

They sell hardware and data management systems but provide limited-to-no insights and are hard to plug into other solutions. They are moving quite slowly in this AI-powered direction, and have not yet taken advantage of recent advances in AI in an environment that is well-suited for an AI approach. Greenhouses can control most inputs (including temperature, water, light, and nutrients), and there are more and more data sources available for measuring output (including computer vision systems and various sensors, in addition to harvested yields).

What does Koidra do

Led by Kenneth Tran, Koidra has shown good proof of its autonomous greenhouse technology by winning twice the Autonomous Greenhouse Challenge. Koidra targets to transform the controlled environment agriculture (CEA) by combining aspects of modern artificial intelligence (AI), industrial internet of things (IIoT) technologies, and human intelligence — what we call artificial intelligence of things (AIoT).

Artificial intelligence of things (AIoT) functionality exists at the intersection of traditional and new sciences and technologies, and it enables an improved partnership between human intelligence and AI. (source)

Koidra’s goal is to empower greenhouses with intelligent autonomous growing technology. Three steps are needed to achieve the final goal:

  • IoT Suite
  • Digital Twin
  • Autonomous growing
A visualization of a deployment roadmap for an autonomous growing solution. This approach may differ for different suppliers.
  1. IoT Suite: The first stage is to connect all the environmental and operational data. Koidra connects data from multiple industry-grade sensors so that there are no data silos. All data is stored and managed in the cloud and can be visualized in a customizable dashboard.
  2. Digital Twin: The second step is to create a “digital twin” based on the real-work data. With the digital twin, growers are able to use AI and machine learning technologies to predict growth, production, and other biological indicators.
  3. Autonomous growing: This is the most exciting part. At this stage, the system can automate decision-making based on historical and live data (e.g. adjust light, water, CO2, humidity) to control the climate and crop production. As a result, it creates a data-driven AIoT system to solve real-time inefficiencies with minimal manual involvement. Growers now can automatically adjust the environment settings, thus increasing crop yield and production efficiency.

Ωpera (pronounced opera) is Koidra’s operators-driven automation platform. It democratizes industrial automation by enabling the production operators and managers to be in control of their automation strategies.

Furthermore, greenhouse control is one proof of concept of the AIoT applications in the agriculture industry. Besides greenhouses and CEA, Koidra is aiming to boost business efficiency and automate operational decision-making in industrial automation. Powered by AI, Koidra is targeting the trillion-dollar market of industrial IoT and IoT-driven automation.

That’s why we are excited to share our investment in Koidra. Ospraie Ag Science led this Seed round, joined by Foothill Ventures and Cavallo Ventures. Check out the funding announcement here.

Why we invested in Koidra

We believe that Koidra has advanced technology, deep domain knowledge, and superior results. It has built a large-scale data infrastructure with strong ML integration which has demonstrated promising results by twice winning the Autonomous Greenhouse Challenge. It is credible that they outperformed the expert Dutch growers.

Koidra provides unique value-add and innovation compared to unfavorable entrenched market leaders. We believe that Koidra is well poised to challenge current leaders.

We believe Kenneth is a great founder. He has led the team to twice win the Autonomous Greenhouse Challenge. Prior to founding Koidra, he was a research scientist at Microsoft Research and worked on real-world reinforcement learning research with applications in data center energy optimization and controlled environment agriculture. He is a strong ML/AI expert, combined with a deep understanding of agriculture and a strong entrepreneurial spirit.

Koidra shows early but promising results through its traction with several large farms in the US and Canada. We get super stellar feedback from customers

We believe there is a strong cross-border market: part of the team is based in Vietnam and has the potential to enter and lead the greenhouse automation market there. This is also the cross-border playbook that we like (usually it is between US and China, but we are also seeing it in other countries like Vietnam and Turkey)

“We believe IA (Intelligent Automation) is a key driver for increasing efficiency and sustainability moving forward, and we are excited to invest in Koidra as a tech leader in IA for climate control, farm automation, and advanced manufacturing,” comments Eric Rosenblum, Managing Partner at Foothill Ventures.

More agtech deals!

Agtech is becoming more popular with VC funding. Based on Pitchbook data, Agtech is blooming with near-record VC funding in Q1 2022. $3.3B was invested in 222 agtech deals globally in Q1, which represents a 15.5% quarter-over-quarter increase in deal value. And data-driven productivity gains is one of the key factors driving the VC momentum into agtech deals.

We are excited about Koidra and are confident that this will certainly not be our last bet in this sector. We are seeing a lot of innovations happening to modernize the agriculture industry, and constantly looking for interesting opportunities, especially around AI applications and robotics.

If you are working on some interesting projects, or know someone in this field, we would be more than happy to connect and learn more.

About Koidra Inc.

Koidra is an intelligent automation (IA) startup based in Seattle and focused on industrial IoT for agricultural and industrial applications. Koidra was founded by Kenneth Tran, a former principal applied scientist at Microsoft, with a mission to transform controlled-environment agriculture, and sustainable agriculture at large, using artificial intelligence of things technology. For more information, please visit www.koidra.ai.

--

--