“Our software enables growers to spend less time on operational issues and more on tactical decision-making in the greenhouse,” says Kenneth Tran, Founder and CEO of Koidra.
Koidra is a provider of intelligent automation solutions with a mission to optimize operations in greenhouses, vertical farms, wood pellet plants, food processing factories and other industrial settings. The overarching goal is to help various industries meet their production goals while keeping costs low and meeting sustainability goals.
In the greenhouse context, Koidra’s AI program works by combining expert knowledge and data from various greenhouse sensors to identify the crop’s status and adjust the greenhouse environment accordingly, whether this be in terms of humidity, temperature, nutrient levels, irrigation, lighting, etc.
Koidra at the Homegrown Innovation Challenge in 2023
Reinforcement learning a challenge in an everchanging greenhouse environment
As Kenneth explains, Koidra’s product is model-based reinforcement learning. This is different from traditional machine learning which collects data to understand the environment and provides insight without taking action. Reinforcement learning (RL) is the science of decision-making, based on learning the optimal behavior in an environment to obtain the maximum reward.
In a greenhouse environment, this involves using sensors and cameras to recognize the crop’s status and then autonomously adjust the greenhouse environment to optimize crop growth. Reinforcement learning is the basis of the Autonomous Greenhouse Challenge, an initiative organized by WUR that challenges different international teams to grow a crop remotely using AI and agrorobotics. Koidra is a familiar face at the AGC, as the company led the winning team in the third edition and was a member of the second edition’s winning team.
“The further we go, the more challenges we see that need to be solved. That said, this technology is applicable not just for greenhouse growth, but also for industrial applications as climate control is common across many industries,” says Kenneth.
Beyond AGC and into a commercial greenhouse
Following its success in the Autonomous Greenhouse Challenge, Koidra looked to bring its technology to commercial greenhouses and was introduced to Great Lakes Greenhouses, one of the largest producers of cucumbers in North America.
“Building on our record yields for long English cucumbers in the Autonomous Greenhouse Challenge in 2018, we ran a trial in 2022 with Great Lakes Greenhouses in 2022 and helped them improve their mini cucumber yields by 19.6%,” Kenneth explains. “The challenge with new technologies is that we see a lot of new technology but they don’t share data or numbers. At Koidra, we have the most data when it comes to greenhouse cucumbers.”
With Koidra’s model, growers always have the option to switch between manual and autonomous input modes.
“Growers can expect higher productivity when using our software and more autonomy to work on tactical decision-making problems as opposed to other operational micro-control issues,” says Kenneth. “Those real-time operational decisions can generally be quite easy to make and easy to automate, and current technology allows us to pass these decisions off to AI.”
Koidra is eager to continue implementing its AI solution in multiple commercial greenhouses to bring enhanced productivity and autonomy to more and more growers.