0000000000684130
AUTHOR
Sylvain Villette
Aerial multispectral imagery for site specific weed management
AGROSUPSPEEAGESTADCT1DOCT; In an agroecological context (French Law on the Future of Agriculture), the reduction of herbicide uses become a crucial issue. It requires developing new technologies allowing a better knowledge of the field. The company AIRINOV, specialized in use of Unmanned Aerial Vehicle (UAV) dedicated to precision agriculture wants to explore a new approach by UAV to localize weed infestation areas. This work aims to develop a method analyzing images acquired by UAV to detect and map weeds in a field. The study is carried out on crops widely grown in France and requiring weed control at a young state of the plant : maize, sunflower and sugar beet. With its on-board sensors …
Chapitre 10: Le désherbage de précision
International audience; Trop longtemps dépendante des seuls herbicides, l'agriculture doit évoluer vers une gestion des plantes adventices, ou " mauvaises herbes ", s'appuyant sur un ensemble de pratiques culturales et de régulations écologiques, avec l'objectif de limiter l'abondance des adventices, de garantir le revenu des agriculteurs, tout en préservant la biodiversité et la qualité de l'environnement. Cet ouvrage explore les moyens et stratégies de gestion durable des communautés adventices faisant appel aux principes de la protection intégrée et à une démarche " agroécologique ".
L'imagerie multispectrale embarquée pour caractériser la croissance et l'état sanitaire du feuillage de la vigne
Multispectral imaging systems are widely used in remote sensing and applied to viticulture context for the canopy characterization. This technique is not used in proximal sensing, to characterize vineyard foliage. Yet the results of field tests led in fixed position have revealed its capacity to estimate the leaf area. The aim of this project is to assess the suitable of a multispectral imaging system as an embedded sensor for vine foliage characterization. To this end, a multispectral camera acquiring visible and near-infrared images and a Greenseeker RT-100 apparatus providing an NDVI (Normalized Difference Vegetation Index), were installed on a track laying tractor. It was equipped with …
Robotics for weed control: I-Weed Robot for a specific spraying
International audience; To preserve environment for a sustainable agriculture, we explore the development of a new autonomous robot, called I-Weed Robot (Intelligent Weed Robot), which aims at reducing herbicides in crop fields (maize, sunflower...). Using a high precision positioning signal (RTK) to locate the robot in the field, a Kaman filter and a proportional-integral-derivative controller (PID controller) allow adjusting the orientation of the robot depending on a predefined trajectory. As for the spraying system, a camera in front of the mobile platform detects weed plants thanks to an image processing based on a crop/weed discrimination algorithm (Hough Transform). At the back a spr…
Simulateur de pulvérisation localisée d’herbicide
Site-specific spraying is a technological lever to reduce the herbicide amount used in the case of chemical control of weeds. At the present time, detection and control technologies are available to apply herbicide only where weeds are present and thus provide an alternative to homogeneous application of chemicals on the entire surface of the field. In order to assess this practice, a digital simulator has been developed combining virtual infestation maps with different theoretical sprayer configurations. Simulations were used to localize herbicide applications and quantify herbicide reduction based on the spatial distribution of weeds and the width of independent spray units (boom sections…
« On-the-go » multispectral imaging system to characterize the development of vineyard foliage
International audience; In Precision Viticulture, multispectral imaging systems are currently used in remote sensing for vineyard vigor characterization but few are employed in proximal sensing. This work presents the potential of a proximal multispectral imaging system mounted on a track-laying tractor equipped with a Greenseeker RT-100 to provide an NDVI index. The camera acquired visible and near-infrared images which were calibrated in reflectance. Vegetation indices were computed and compared to Greenseeker data. From two of the resulting datasets, a spatio-temporal study of foliage description through both optical systems is presented. This first study assessed the proximal imagery re…
A virtual spreader to overcome experimental limits: Example of use to deepen the meaning of the transverse coefficient of variation
National audience; Nowadays, centrifugal spreading is the most used method to apply fertiliser in fields. Numeroustechnical developments have improved spreaders to reach high precision applications.Nevertheless, these developments have never taken into account the influence of the spreadingvelocity variation on the shape of the spread pattern. However, practical examples show that speedvariations regularly happen in the field. For instance, high speeds applications require punctualslowing down because of:- Trees or posts;- Slopes, hills;- Field borders/ends.Technical devices already exist to avoid the influence of the travel speed on the disc rotation speedand on the application rate in fie…
Towards the estimation of vole damage on grassland by aerial multispectral imaging
International audience
A crop field modeling to simulate agronomic images
In precision agriculture, crop/weed discrimination is often based on image analysis but though several algorithms using spatial information have been proposed, not any has been tested on relevant databases. A simple model that simulates virtual fields is developed to evaluate these algorithms. Virtual fields are made of crops, arranged according to agricultural practices and represented by simple patterns, and weeds that are spatially distributed using a statistical approach. Then, experimental devices using cameras are simulated with a pinhole model. Its ability to characterize the spatial reality is demonstrated through different pairs (real, virtual) of pictures. Two spatial descriptors …
Centrifugal spreading : motion blurred images to determine the three components of fertiliser outlet velocity
In the optical control of centrifugal spreading, the use of a single CCD camera to measure the outlet velocity is not limited to flat discs and 2D motion estimation. Combining the kinetic study of the fertiliser motion on the vane and the geometric analysis of image acquisition, the 3D components of the outlet velocity can be deduced from motion blurred images in the case of a traditional concave disc. The estimation of the horizontal and vertical outlet velocities is useful to predict spread patterns using ballistic flight models. This opens up the possibility to implement simple spreader test tools for quality diagnoses or sensors for feedback loop adjustments.
Détection non supervisée des adventices par drone : résultats et limites
Dans un cadre de diminution des produits phytosanitaires, l'agriculture de précision est une solution technique pour diminuer l'impact environnemental de l'agriculture sans transformer les systèmes de production actuels. La démocratisation des drones aériens pour l'agriculture permet leur utilisation afin de discriminer culture et adventices au sein de parcelles cultivées. Nous avons développé et testé des algorithmes non supervisés (ne nécessitant pas l’intervention d’un humain) combinant l'information spatiale et spectrale pour réaliser cette discrimination. Cette présentation sera l'occasion de revenir sur les résultats de ces algorithmes et de présenter également les limites rencontrées…
I-Weed robot : un outil pour l'étude de population de plantes adventices
The I-Weed robot is a small autonomous platform that aims to characterize the vegetation in an inter-row maize crop for a spot spraying: spraying is done only where a weed is detected. This article presents how the robot works and aims to experimentally evaluate the ability to map weeds from a WeedSeeker system (Trimble) embedded on the I-Weed robot. The results are compared to those from an imaging system, used as ground truth. After analyzing a confusion matrix, it reveals that the system is able to correctly geotag plants at 75 % and is ready for a spot spraying.
Aerial imagery for site specific weed management
National audience
Weed detection by aerial imaging: impact of soil, crop and weed spectral mixing
International audience; This study aims to evaluate spectral information potential of images captured with a UAV, for site specific weed management. The image acquisition chain was modeled in order to compute the digital values of image pixels, according to the field conditions and objects lying on the ground surface projected in the pixels. The object spectra are mixed in the same pixel to estimate the impact of the spatial resolution of the image. The classification potential into crop, weed and soil classes was studied usinf simulations based on the present multispectral sensor characteristics and according to different mixing rates.
I-Weed Robot : un robot autoguidé pour un désherbage localisé
We present the development of an autonomous robot, called I-Weed Robot (Intelligent Weed Robot), which aims at reducing herbicides in crop fields (maize, sunflower…). Using a high precision positioning signal (RTK) to locate the robot in the field, a Kalman filter and a proportional-integral-derivative controller (PID controller) allows adjusting the orientation and the speed of the robot depending on a predefined trajectory. As for the specific spraying system, a commercial system is used (weedseeker, Trimble) where the plant detection is obtained by an optical sensors just before to spray specifically on them. The performance of the guidance algorithm using numerical simulations (virtual …