Thesis subject

Use Reinforcement Learning to Control Greenhouse

Greenhouse is an important approach in modern agriculture. However, it is difficult to control the greenhouse environment, such as temperature, humidity, light intensity, and carbon dioxide concentration, by the classical methods due to uncertain nonlinear control. Therefore, it is urgent to develop an intelligent control framework to control the setting points of the greenhouse using artificial intelligence solutions.

In this project, we aim to use reinforcement learning (RL) for greenhouse environment control. Specifically, sensors and cameras are used to recognize the growing states of crops, and actuators are used to control the irrigation system, heating system, and ventilation system. The aim of the reinforcement learning model is to find the optimal control solution for the greenhouse.

background
Specifically, this project includes the following tasks:

  • Identify variables of interest within the farm environment.
  • Build a simulation environment of a target greenhouse.
  • Use reinforcement learning to calculate the optimal control solution for the simulation greenhouse.

The content above is an overall description of the project. The detailed research work of the project could be based on further discussion between supervisors and students.

Literature

Requirements (optional)

  • Pyhon programming
  • Machine learning

Contact person

  • qingzhi.liu@wur.nl