Koidra
3 Case Studies
A Koidra Case Study
Microsoft Research, a leading technology research organization, was seeking ways to apply artificial intelligence to address global food production challenges. They faced the specific challenge of using AI to remotely maximize cucumber yield in a greenhouse while minimizing resource usage for the 2018 Autonomous Greenhouse Challenge. The vendor, Koidra, through its team's expertise in reinforcement learning, provided the AI-driven solution for this controlled environment agriculture project.
Koidra implemented a solution using an approximate Bayesian model-based reinforcement learning framework. This AI agent was designed to control greenhouse settings like lighting, temperature, and nutrients remotely. The results were highly successful; Koidra's solution produced 55 kilograms of cucumber per square meter with a net profit of €25/m², beating both expert human growers and other competing AI teams to win the challenge. This demonstrated a significant impact, achieving a 17% higher net profit than the manual reference and a 25% higher net profit than the second-place team.