We help you find the perfect fit.

Swiss Ai Research Overview Platform

SMart Agriculture using Lasers and Artificial intelligence

Lay summary

La réduction de l’usage de pesticides dans l’agriculture est un enjeu essentiel dans nos sociétés. Mais la santé des cultures l’est tout autant, ne serait-ce que pour assurer l’alimentation des populations. Une approche raisonnable, réconciliant ces deux impératifs, est de ne traiter les cultures par des produits phytosanitaires que quand et le risque d’infection est véritablement avéré et significatif. Ceci implique une mesure des paramètres pertinents, comme la présence de spores pathogènes et les conditions environnementales, sur de grandes surfaces mais à haute résolution spatiale et en temps réel.

Le but du projet SMALA est de répondre à ce besoin en déployant des réseaux de stations laser (IoT) capables de mesurer ces paramètres (développées par l’Université de Genève) et les intégrer dans des modèles prédictifs de risque existants (par l’Agroscope à Changins) afin d’en avertir les producteurs. La méthode de détection des spores pathogènes se fait par une méthode originale brevetée basée sur l’holographie laser et un traitement des données par intelligence artificielle (AI).

La culture principale sur laquelle le projet de focalise est la vigne, dont les pathogènes considérés sont l’oidium et le mildiou, mais la méthode sera également étendue à d’autres cultures.

Abstract

The reduction of pesticide use in agriculture has become a major socio economic and environmental concern. The project that we propose is dedicated to the development of novel approaches and technologies, involving photonics (lensless microscopy, ptychography, wide field digital holography), Internet of Things (IoT), artificial intelligence and modeling in order to treat crops only when and where the risk of infection is high, and early enough to avoid plant disease and further pathogen spread in the fields. These methods will provide high spatial (sub-field) and temporal resolution (real time) measurements, and assess the mechanisms of the primary infection and the distances over which pathogen spores are transported are also merely unknown. The time and spatially resolved spore concentration measurement will be incorporated in the Agrometeo/Vitimeteo predictive models from the Agroscope. After validation of the method, a service will be proposed to the farmers in order to optimize pesticide sprays, and thus reduce both the environmental impact and their production costs. The implementation of the project, based on the unique expertise of the UniGe on bioaerosol identification and of the Agroscope on agronomy and pesticide reduction modeling, will be performed as follows:- Development of novel lensless IoT microscopes based on laser wide field digital holography, and ptychography- Create databases for downy and powdery mildew spore, using reference samples from the Agroscope fungal library, including their physiological variations as a function of age, environmental conditions, strains; create databases of the most frequent interfering spores found in vineyards- Adapt and optimize deep learning based image recognition algorithms using the spore databases- Design, build, and test in the field (Agroscope Changins and Pully) a network of stand-alone, low-cost lensless IoT microscopes, with autonomous solar power supply. - Assess the domains of applications like the study of initial infection mechanisms and of the infection spread in vineyards by spore transport- Couple the high-temporal resolution spore identification at each measurement location with the weather data, to improve the biological models underlying the Vitimeteo/Agrometeo forecast.- Implement real-time incorporation of the spore measurements into the Vitimeteo/Agrometeo model.- Determine of intervention threshold in situ. The availability of real-time quantitative spore measurements, together with improved disease development and pressure modelling, will provide unprecedented calibrated data. This quantitative information, compared with scouting information about disease development in untreated plots, will be used to develop optimized fungicide spraying program, maximizing efficiency while minimizing pesticide use.The project will concentrate on two pathogens in vineyards (downy mildew and powdery mildew) but extension to other agriculture sectors will be considered as well.

Jean-Pierre Wolf
  Jean-Pierre Wolf
Pierre-Henri Dubuis