This concept paper presents Mobile Agricultural Robots (MARs) for the development of precision agriculture and implicitly the smart farms through knowledge, reason, technology, interaction, learning and validation. Finding new strategies and control algorithms for MARs has led to the design of an Autonomous Robotic Platform Weed Control (ARoPWeC). The paradigm of this concept is based on the integration of intelligent agricultural subsystems into mobile robotic platforms. For maintenance activities in case of hoeing crops (corn, potatoes, vegetables, vineyards), ARoPWeC benefits from the automatic guidance subsystem and spectral analysis subsystem for differentiation and classification of the weeds. The elimination of weeds and pests is done through the Drop-on-Demand spray subsystem with multi-objective control, and for increasing efficiency through the Deep Learning subsystem.