Abstract Presently, due to the social demand, two trends weigh heavily on the agricultural and food production : foodstuff quality enhancement and the search for more environmental-friendly agricultural practices. Agriculture and the food industry are expected to take into account technical, social, economical and environmental factors. To do this means developing adaptability, and to adapt quickly one requires reliable and accurate information. This is why it is developed technologies and methodologies for acquiring and processing information in the agricultural fields and in the food processing plants. In agriculture, our main goals are to collect information from the field and to accurately model the cultivation practices that need to be improved. Embedded sensors are based on artificial vision and microwaves. In artificial vision, the main obstacles to be overcome are heterogenous lighting conditions, the great variability of biological products and the segmentation process (due to overlapping objects). Complex cultivation practices such as pesticide spraying or manuring are either optimised — using fluid mechanics for spraying, balistics for manuring- or replaced by robotic-based techniques. In food processing, research has been carried out on the modelling of the human perception and decision-making when assessing food quality. Powerful investigation techniques such as magnetic nuclear resonance are currently being studied to provide more in-depth knowledge of food products. Low—cost sensors (based on artificial vision, infrared spectroscopy) are developed to allow on-line, non destructive measurement of food characteristics (for instance sugar content in fruit). That implies technology studies (choice of the components, design), modelling (for instance, the light distribution in fruit), and chemometrics. Finally, the decisions of the human operator of food processing lines are modelled by artificial intelligence : fuzzy rules are created using either expert knowledge or fuzzy inference systems. These research methods are consistently applied to relevant and every-day problems. They engender multidisciplinary approaches carried out under heavy constraints (of cost, real-time, robustness) at the crossroads where Engineering, and Biology meet Social demands.