What is AI? How will it be used in greenhouses, indoor farms and commercial horticulture?

The term “AI” or artificial intelligence is increasingly being discussed in all facets of life. This is true for commercial horticulture as well.

Whether you grow ornamental crops, cannabis or greenhouse vegetables there are numerous companies that are promising to optimize processes, increase efficiency, and improve overall crop yields.

But, what are they really doing? Does it work? Will it work? Should I invest?

After having many conversations with growers in greenhouses, vertical farms and other indoor (and outdoor) growing facilities, we decided to put together a short article outlining the types of technologies being proposed and their capabilities at this stage in the product development process.

The way we see it, there are five types of “AI” terminologies/technologies being marketed, developed and used in some form or fashion: Digital Twins, Generative AI, Intelligent Algorithms, Sensing/Computing Vision and Robotic.

Digital Twins: A digital twin is a virtual model of a physical object. It spans the object’s lifecycle and uses real-time data sent from sensors on the object to simulate behavior and monitor operations.	Generative AI: Generative AI focuses on creating new and original content, chat responses, designs, synthetic data or even deep fakes. It’s particularly valuable in creative fields and for novel problem-solving, as it can autonomously generate many types of new outputs.
Intelligent/Smart Algorithms: Intelligent algorithms are, in many cases, practical alternative techniques for tackling and solving a variety of challenging engineering problems.	Sensing/Computer Vision: Computer vision focuses on enabling computers to identify and understand objects and people in images and videos. Like other types of AI, computer vision seeks to perform and automate tasks that replicate human capabilities.
Robotics: Robotics is both a platform and an application. Robotics may pull from the other categories, ultimately acting as the hands of AI. It may be what provides a feedback loop though computer vision or it may complete the task created by the intelligent algorithms.

Based on the research we have access to, including websites of businesses targeting commercial horticulture and controlled environment agriculture, we see companies using these five types of AI to enhance the following functions:

Climate Control and Monitoring:
- Analyze and interpret data from various sensors that monitor temperature, humidity, light, and CO2 levels within the greenhouse.
- Based on this data, proprietary algorithms make real-time adjustments to climate control systems, such as adjusting ventilation, heating, and cooling, to create optimal growing conditions for vegetables.
Precision Irrigation:
- Aid in precision agriculture by analyzing data related to soil/soilless media temperature, nutrient levels, and water usage.
- This information is used to optimize irrigation schedules, ensuring that plants receive the right amount of water and nutrients, thereby reducing waste and improving resource efficiency.
Crop Monitoring and Disease Detection:
- Cameras and sensors can monitor the health of plants by analyzing images and detecting signs of diseases, pests, or nutrient deficiencies.
- Early detection allows for prompt intervention, reducing the use of pesticides and minimizing crop losses.
Predictive Analytics:
- Algorithms can analyze historical data and environmental conditions to predict future trends and optimal planting times.
- Farmers can use these predictions to plan their planting schedules, ensuring a consistent and efficient production cycle.
Harvesting Automation:
- Robotics and robotic systems can be employed for automated harvesting of vegetables. These robots use computer vision to identify ripe fruits and vegetables, ensuring accurate and efficient harvesting.
Sorting and Grading:
- Cameras and sensors are used for automated sorting and grading of harvested vegetables based on quality, size, and other parameters.
- This automation streamlines the post-harvest process, improving efficiency and reducing labor requirements.
Data-Driven Decision Making:
- Advanced computing and modeling systems analyze large datasets to provide insights and recommendations for optimizing various aspects of greenhouse vegetable production.
- Farmers can make informed decisions on factors such as planting density, crop rotation, and resource allocation.
Energy Management:
- Advanced algorithms help to optimize energy usage within the greenhouse by analyzing patterns and adjusting lighting, heating, and cooling systems for energy efficiency.
Supply Chain Optimization:
- Advanced algorithms can be used to optimize the supply chain by predicting demand, managing inventory, and ensuring timely delivery of greenhouse vegetables to market.

Does AI enhance these functions?

The short answer is yes. Based on industry-supported competitions and ongoing trials at commercial growing facilities as well as research institutes, growing evidence supports using AI in your controlled environment agriculture facility.

The long answer is, it’s complicated. The reality is, we are still new in this journey of automating farms. All crops and farms are different or think they are different. They have varying configurations, varying degrees of usable data with different climate management systems, sensors, cultures, security requirements and operating systems.

Today’s farms were built over decades and have a wide variety of technology in them (many times at the same location). They take drastically different approaches to storing and managing data. And, most importantly, they have vastly different budgets.

Assuming the statement “many farms are different” is true, it will take time, partnerships and patience for engineers to gain access to enough farms with good data to make new technologies work at scale. No one company can offer all these solutions any time in the near future.

What’s likely is that your climate control computer system provider will need to work with sensors provided by multiple companies. Each of these companies will need to learn how to work together with companies that develop algorithms. To add another layer of complexity, they will then need to work with equipment and component suppliers to make sure the equipment can communicate within the complex ecosystem and environment created by greenhouse operations over the past several decades.

But (there is always a “but”)…

Based on data from projects and organizations such as Ohio State University, Rutgers, the University of Arizona, Wageningen University (WUR), the Autonomous Greenhouse Challenge, AGROS, Delphy and numerous grower trials, mounting evidence supports the use of climate control and monitoring, data-driven decision making, energy management and supply chain optimization, primary driven by intelligent/smart algorithms and, in a few cases, generative AI. Farms that are ideal candidates at this time are new high-tech greenhouses producing one crop. The best data currently available is on cucumbers, tomatoes and leafy greens.

So should you invest time and energy into AI now?

Yes is the easier answer. At this stage, your budget, ability to collect data, share data and farm’s culture of technology adoption should determine how much time, money and energy you should invest. It’s safe to say that every farm, at some level, will have more positive impacts than negative ones.

Our recommendation is that you start thinking about two factors:

What pain points could you alleviate with AI?
What data do you need to collect to allow AI to work for you?

Integrating AI in greenhouse vegetable production or other commercial horticulture facilities may contribute to sustainable farming practices, resource conservation, and increased productivity. Much of this is still not known or proven. What is known for sure is that successful implementation will require collaboration between agricultural experts, data scientists, and technology developers. Artificial intelligence as it stands today has to learn from data and statistics available to it. In other words it is not going to solve problems to answers not yet known. Advanced algorithms, however, can take proven science and help growers and farmers process more data than ever possible before.