Cannabis growers are looking closely at mainstream agricultural technology to gain a new level of precision cultivation that employs everything from artificial intelligence and augmented reality to robots and drones.

The goals: better-quality cannabis, lower labor costs and improved productivity.

Over the past few years, a host of AI-powered solutions have emerged to address:

• Cannabis growing methods, or protocols.
• Day-to-day labor savings.
• Growth-rate tracking.
• Yield forecasting.
• Inventory management.
• Profitability management.

These are typically digital sensor-based systems managed by various platforms – Aroya, Canix and Trym are a few examples – that control lighting timing, lighting intensity, irrigation, temperature, humidity, fans and airflow, CO₂ levels and plant nutrients.

“If you’re looking at how cannabis operations are run, everyone is doing their own trials or experiments already,” said Ben Niehaus, founder of Dresden, Germany-based SpexAI.

SpexAI is developing and testing a smart camera, called Hugin.

Mounted on a robot, the camera rolls down aisles in a greenhouse at night scanning the plants and then combines the camera’s spectral imaging with machine learning and artificial intelligence to detect diseases and cannabinoids directly from the growing plants.

The technology creates a 3D model of each plant using 1,000 data points, then combines that with environmental data to see how the plant is progressing.

“The growers are always trying to optimize the plant, and they have new genetics coming in,” Niehaus said.

“They try to find the best strategy. So, in a sense, they are already researchers.”

Growth of digital horticulture

 One relative newcomer to cannabis tech is AgriSmart Engineering, based in South Africa.

According to AgriSmart CEO Constant Beckerling, the company focuses on three areas:

• Recirculating hydroponic systems, where water is continuously recaptured and recirculated to the plants, with nutrients injected as needed.
• LED lighting.
• Data management and automation.

“We combine these three different skill sets and technology streams into a turnkey facility design, where we work with commercial cultivators or prospective commercial cultivators to set up cultivation environments, which are efficient but also compliant,” Beckerling said during a phone interview with MJBizDaily.

AgriSmart has set up a turnkey digital system in a hydroponic cannabis cultivation facility in Johannesburg.

The company has identified critical areas where technology such as AI can play a crucial role, Beckerling said, including in a supplemental lighting context with a smart, reactive dimmer.

“A sensor would tell you what incident solar radiation would be, and based on that, AI could brighten or dim the plant’s LED light automatically,” he said.

“You could look at the nutrient composition of your plant tissue, then use AI to plan irrigation schedules and feeding schedules.”

Growing AI influence

Justin Clune, CEO of Skysense, a drone imaging and AI AgTech company in San Mateo, California, saw a problem in his own cultivation operation: male and female cannabis plants mixed up in a grow.

“We had about one full-time laborer for every 10 acres just to walk between the rows of the plants and look at each plant to see if they are male or female,” Clune said.

“It’s a needle in a haystack problem. We would basically double-check the same place twice a week for up to eight weeks.”

So he thought of having a drone carrying a sensor do the scouting – in other words, one drone could do the scouting of 10-20 people on the ground.

“The hard part is not the idea. It’s always the execution,” Clune said.

He heard from others who had tried the idea, some claiming that an expensive multispectral sensor was needed to determine the sex of the plant.

“Everyone thought that there’d be like this one sensor type, just this silver bullet. And that’s just not true,” Clune said.

What he found that works is something that emulates how humans work.

AI essentially imparts to the drone the same sort of domain-specific knowledge that a farm manager or field worker has, Clune said.

“AI is really the best technology for the job,” he added.

The future vision

 “From my perspective, controlled environment agriculture (CEA) was just a control problem. It’s not an agricultural problem,” said Kenneth Tran, CEO and chief technology officer of Seattle-based intelligent automation startup Koidra.

Tran determined that one way to handle greenhouse systems engineering in advance is to create a 3D greenhouse model.

“Our greenhouse digital twin model includes two intertwined components,” Tran said.

“One is a crop model, and one is a greenhouse climates model.”

Tran said the company has crop models for tomatoes as well as lettuce and leafy greens – but not yet for cannabis.

“It’s just a matter of time,” he said.

“If someone asked for it, and we could develop the crop model for cannabis, then we just plug that crop model into the greenhouse model, which is the greenhouse digital twin.”

The digital twin then informs the optimization engine – which has been trained through a machine learning process – to take the best actions on a real-time basis, such as increase or decrease energy usage, add or curtail irrigation or whatever the goal is to optimize profitability.

A role for augmented reality?

Augmented reality (AR) for cannabis cultivation could also be on the horizon.

Joseph Peller, a spectral imaging professor at the Wageningen University & Research in the Netherlands, ranked as the world’s top institution for agriculture science, is part of a team wrapping up a four-year project using an augmented reality headset on crops.

His team demonstrated that it is possible to detect ripeness characteristics of tomatoes using the 3D scanning capabilities of the company’s HoloLens AR device.

“We verified this in an experimental setup with multiple tomato varieties,” he wrote in a 2022 study.

Peller noted there have been research efforts with cannabis at Wageningen, such as 3D scanning of the tops of plants to estimate yield.

There might be an upcoming opportunity to use AR technology in cannabis production.

“But I don’t think I can speak about the specifics about it because of the weird legalities,” Peller wrote.

The cannabis grow of the future

 Seven to 15 types of hardware/software subsystems – such as HVAC control systems and electricity-usage systems – run the average large indoor grow facility, according to Vince Harkiewicz, co-founder and CEO of Grownetics in Boulder, Colorado, which developed a fully integrated sensing and automation platform for cannabis cultivation.

“If you truly wanted autonomy, you’d have to integrate every single subsystem,” Harkiewicz said. “And it just hasn’t been done yet.

“The position we’re taking is to look for all the systems and new systems that are coming out in the future. To even be able to pull the data from them to apply and teach these AI algorithms, we need a middleware that ties it all together.

“And that’s what we set out to build.”

The Grownetics middleware is a time-series database that acts as a sort of universal translator, he said, by aggregating all the data.

The technology then spits out reports and recommendations based on that data from the various sensors in the subsystems gathered during operations – such as HVAC-usage readings, lighting levels, soil moisture, temperature, relative humidity, CO₂ levels, irrigation water amounts, levels of electricity used throughout the day.

“The goal of autonomous cannabis cultivation is to steer plant growth characteristics by short circuiting and manipulating all the stimulus the plant would have received in the natural environment with the ones we create and control down to the millisecond,” Harkiewicz told MJBizDaily via email.

“CEA is a unique type of bioengineering as we are truly creating alien environments that plants would never experience in nature to accelerate or direct them to produce the valuable compounds we’re looking for.”

Source: https://mjbizdaily.com/cannabis-growers-eye-digital-tools-artificial-intelligence-to-improve-crops-productivity/