In this project the main goal is to close the interaction loop between 3 actors: robotic units, living plants and human stakeholders. By employing specific algorithms of collective-level control, a grid of robotic agents influences the direction of growth of plants, while they simultaneously sense the growth of the plant, in order to react to it. Sensing of plants is either done by proximate IR sensing or by computer vision based on cameras. The influence onto the plant growth is achieved by emitting light of specific wave lengths (red, blue or far-IR) and by coordinating these emissions with other robot nodes. Which patterns of stimulus emissions create the appropriate settings to let the plants grow in a desired manner is explored by machine-learning algorithms, creating a closed loop system that can run autonomously, with no human engineer, with no human designer and with no human programmer needed to be part of the loop.