The recent developments in agriculture, in particular hydroponic cultivation techniques, represent one of the key areas of applications of IoT and Information and Communication technologies in order to achieve improved efficiency from quantitative analysis (data-driven) not only of the entire crop but also at the level of each individual plant (plant driven approaches). Thanks to sensors and machine learning algorithms, it is possible to gather information on the state of health of individual seedlings, fruits controlling the injection of nutrients, water and plat protection products within hydroponic greenhouse taking also into account energy efficiency in the design of new control loops for automation in this domain. In this article, we present the architectural proposal of a robotic handling system designed to optimize the positioning of the seeds inside a domestic hydroponic greenhouse, aiming to achieve an automatic and more precise seedlings process. The main contribution of the article is to highlight the relevance on obtaining the optimal positioning of seeds for future growth of plants in the controlled environment. In our case, tomato seeds were considered with the aim of obtaining a homogeneous growth of all the seedlings in the context of a highly optimized artificial environment composed by sensors, actuators, LED lighting, etc and where artificial intelligence techniques will be used for the detection, counting, identifying of tomato seedlings, estimating their precise position in the plant germinator. The architecture involves the use of a robotic arm and a video camera with embedded hardware acceleration used for deep neural network classification. The system processes the data received from the camera and then activates the robotic arm to manage the placement of the seedlings. We will also highlight the need for optimal positioning algorithms and path management to reduce the energy consumption of the system.