Intelligent sensing for production of high-value crops Scientific and technical quality This thesis has been realized within the CROPS project. CROPS will develop scientific know-how for a highly configurable, modular and clever carrier platform that includes modular parallel manipulators and intelligent tools (sensors, algorithms, sprayers, grippers) that can be easily installed onto the carrier and are capable of adapting to new tasks and conditions. Several technological demonstrators will be developed for high value crops like greenhouse vegetables, fruits in orchards, and grapes for premium wines. The CROPS robotic platform will be capable of site-specific spraying (targets spray only towards foliage and selective targets) and selective harvesting of fruit (detects the fruit, determines its ripeness, moves towards the fruit, grasps it and softly detaches it). Another objective of CROPS is to develop techniques for reliable detection and classification of obstacles and other objects to enable successful autonomous navigation and operation in plantations and forests. The agricultural and forestry applications share many research areas, primarily regarding sensing and learning capabilities. The project started in October 2010 and will run for 48 month. The aim of this thesis is to lay the foundations, suggesting the guidelines, of one task addressed by the CROPS project, in particular, the aim of this work is to study the application of a VIS-NIR imaging approach (intelligent sensing), based on a relatively simple algorithm, to detect symptoms of powdery mildew and downy mildew disease at early stages of infection (sustainable production of high-value crops). Also a preliminary work for botrytis detection will be shown. Concept and objectives Many site-specific agricultural and forestry tasks, such as cultivating, transplanting, spraying, trimming, selective harvesting, and transportation, could be performed more efficiently if carried out by robotic systems. However, to date, agriculture and forestry robots are still not available, partly due to the complex, and often contradictory, demands for developing such systems. On the one hand, agro-forestry robots must be of reasonable cost, but on the other, they must be able to deal with complex, dynamic, and partly changing tasks. Addressing problems such as continuously changing conditions (e.g., rain and illumination), high variability in both the products (size, and shape) and the environment (location and soil properties), the delicate nature of the products, and hostile environmental conditions (e.g. dust, dirt, extreme temperature and humidity) requires advanced sensing, manipulation, and control. Since it is impossible to model a-priori all environments and task conditions, the robot must be able to learn new tasks and new working conditions. The solution to these demands lies in a modular and configurable design that will keep costs to a minimum by applying a basic configuration to a range of agricultural applications. At least a 95% yield rate is necessary for economical feasibility of an agro-forestry robotic system. Objectives An objective of CROPS project is to develop an “intelligent tools” (sensors, algorithms, sprayers) that can easily be installed onto a modular and clever carrier platform. The CROPS robotic platform will be capable of site-specific spraying (targeted spraying only on foliage and selected targets). Research efforts To achieve the novel systems described above, we will focus on intelligent sensing of disease detection on crop canopy (investigating different types and/or multiple sensors with decision making models). Technology evaluation Technology evaluation of the developed systems will include the performance evaluation of the different components (e.g., capacities, success rates/misses). Progress beyond the state-of-the-art Despite the extensive research conducted to date in applying robots to a variety of agriculture and forestry tasks (e.g., transplanting, spraying, trimming, selective harvesting), limited operating efficiencies (speeds, success rates) and lack of economic justification have severely limited commercialization. The few commercial autonomous agriculture and forestry robots that are available on the market include a cow milking robot, a robot for cutting roses (RomboMatic), and various remote-controlled forest harvesters. These robots either have a low level of autonomy or are able to perform only simple operations in structured and static environments (e.g. dairy farms and plant breeding facilities). Developing capabilities for robots operating in unstructured outdoor environments or dealing with the highly variable objects that exist in agriculture and forestry is still open-ended, and one of CROPS aims is to address this problem. Current state-of-the-art Field trials have routinely shown that most crop damage due to diseases and pests can be efficiently controlled when treatments are applied timely and accurately by hand to susceptible targets (i.e., by intelligent spraying). Site-specific spraying targeted solely to trees and/or to infected areas can reduce pesticide use by 20–40%. An issue of relevance to targeted agriculture is the detection of diseases in field crops. Since such events often have a visual manifestation, state-of-the-art methods for achieving this goal include fluorescence imaging or the analysis of spectral reflectance in carefully selected spectral bands. While reports of these methods used separately achieved performance at 75–90% accuracy, attempts to combine them have boosted disease discrimination accuracy to 95%. We must note here, however, that despite these promising results, very little research has been conducted on in-field disease detection. Expected progress The diseased detection approach for precision pesticide spraying will be developed investigating image processing techniques (after a laboratory spectral evaluation and greenhouse testing) for high-precision close-range targeted spraying to selectively and precisely apply chemicals solely to targets susceptible to specific diseases/pests, with a mean 90% success rate. Local changes in spectral reflection of parts of the canopy will be used as an indication of disease. “Soft-sensor” for detection of ripeness and diseases (noncontact rapid sensing system) will be developed by multispectral sensor (multispectral spectral camera). These “soft sensor” can be used as a decision model for targeted spraying.

ADAPTIVE PROCESSING ARCHITECTURE OF MULTISENSOR SIGNALS FOR LOW-IMPACT TREATMENTS OF PLANT DISEASES / P. Tirelli ; tutor: N. A. Borghese ; correlatore: N.A. Borghese ; direttore: E. Damiani. DIPARTIMENTO DI INFORMATICA, 2013 Feb 26. 24. ciclo, Anno Accademico 2011. [10.13130/tirelli-paolo_phd2013-02-26].

ADAPTIVE PROCESSING ARCHITECTURE OF MULTISENSOR SIGNALS FOR LOW-IMPACT TREATMENTS OF PLANT DISEASES.

P. Tirelli
2013

Abstract

Intelligent sensing for production of high-value crops Scientific and technical quality This thesis has been realized within the CROPS project. CROPS will develop scientific know-how for a highly configurable, modular and clever carrier platform that includes modular parallel manipulators and intelligent tools (sensors, algorithms, sprayers, grippers) that can be easily installed onto the carrier and are capable of adapting to new tasks and conditions. Several technological demonstrators will be developed for high value crops like greenhouse vegetables, fruits in orchards, and grapes for premium wines. The CROPS robotic platform will be capable of site-specific spraying (targets spray only towards foliage and selective targets) and selective harvesting of fruit (detects the fruit, determines its ripeness, moves towards the fruit, grasps it and softly detaches it). Another objective of CROPS is to develop techniques for reliable detection and classification of obstacles and other objects to enable successful autonomous navigation and operation in plantations and forests. The agricultural and forestry applications share many research areas, primarily regarding sensing and learning capabilities. The project started in October 2010 and will run for 48 month. The aim of this thesis is to lay the foundations, suggesting the guidelines, of one task addressed by the CROPS project, in particular, the aim of this work is to study the application of a VIS-NIR imaging approach (intelligent sensing), based on a relatively simple algorithm, to detect symptoms of powdery mildew and downy mildew disease at early stages of infection (sustainable production of high-value crops). Also a preliminary work for botrytis detection will be shown. Concept and objectives Many site-specific agricultural and forestry tasks, such as cultivating, transplanting, spraying, trimming, selective harvesting, and transportation, could be performed more efficiently if carried out by robotic systems. However, to date, agriculture and forestry robots are still not available, partly due to the complex, and often contradictory, demands for developing such systems. On the one hand, agro-forestry robots must be of reasonable cost, but on the other, they must be able to deal with complex, dynamic, and partly changing tasks. Addressing problems such as continuously changing conditions (e.g., rain and illumination), high variability in both the products (size, and shape) and the environment (location and soil properties), the delicate nature of the products, and hostile environmental conditions (e.g. dust, dirt, extreme temperature and humidity) requires advanced sensing, manipulation, and control. Since it is impossible to model a-priori all environments and task conditions, the robot must be able to learn new tasks and new working conditions. The solution to these demands lies in a modular and configurable design that will keep costs to a minimum by applying a basic configuration to a range of agricultural applications. At least a 95% yield rate is necessary for economical feasibility of an agro-forestry robotic system. Objectives An objective of CROPS project is to develop an “intelligent tools” (sensors, algorithms, sprayers) that can easily be installed onto a modular and clever carrier platform. The CROPS robotic platform will be capable of site-specific spraying (targeted spraying only on foliage and selected targets). Research efforts To achieve the novel systems described above, we will focus on intelligent sensing of disease detection on crop canopy (investigating different types and/or multiple sensors with decision making models). Technology evaluation Technology evaluation of the developed systems will include the performance evaluation of the different components (e.g., capacities, success rates/misses). Progress beyond the state-of-the-art Despite the extensive research conducted to date in applying robots to a variety of agriculture and forestry tasks (e.g., transplanting, spraying, trimming, selective harvesting), limited operating efficiencies (speeds, success rates) and lack of economic justification have severely limited commercialization. The few commercial autonomous agriculture and forestry robots that are available on the market include a cow milking robot, a robot for cutting roses (RomboMatic), and various remote-controlled forest harvesters. These robots either have a low level of autonomy or are able to perform only simple operations in structured and static environments (e.g. dairy farms and plant breeding facilities). Developing capabilities for robots operating in unstructured outdoor environments or dealing with the highly variable objects that exist in agriculture and forestry is still open-ended, and one of CROPS aims is to address this problem. Current state-of-the-art Field trials have routinely shown that most crop damage due to diseases and pests can be efficiently controlled when treatments are applied timely and accurately by hand to susceptible targets (i.e., by intelligent spraying). Site-specific spraying targeted solely to trees and/or to infected areas can reduce pesticide use by 20–40%. An issue of relevance to targeted agriculture is the detection of diseases in field crops. Since such events often have a visual manifestation, state-of-the-art methods for achieving this goal include fluorescence imaging or the analysis of spectral reflectance in carefully selected spectral bands. While reports of these methods used separately achieved performance at 75–90% accuracy, attempts to combine them have boosted disease discrimination accuracy to 95%. We must note here, however, that despite these promising results, very little research has been conducted on in-field disease detection. Expected progress The diseased detection approach for precision pesticide spraying will be developed investigating image processing techniques (after a laboratory spectral evaluation and greenhouse testing) for high-precision close-range targeted spraying to selectively and precisely apply chemicals solely to targets susceptible to specific diseases/pests, with a mean 90% success rate. Local changes in spectral reflection of parts of the canopy will be used as an indication of disease. “Soft-sensor” for detection of ripeness and diseases (noncontact rapid sensing system) will be developed by multispectral sensor (multispectral spectral camera). These “soft sensor” can be used as a decision model for targeted spraying.
26-feb-2013
Settore INF/01 - Informatica
Settore AGR/09 - Meccanica Agraria
multi-sensor signals ; precision spraying ; image processing ; grapevine ; powdery mildew ; downy mildew ; botrytis ; clustering ; early detection
BORGHESE, NUNZIO ALBERTO
DAMIANI, ERNESTO
Doctoral Thesis
ADAPTIVE PROCESSING ARCHITECTURE OF MULTISENSOR SIGNALS FOR LOW-IMPACT TREATMENTS OF PLANT DISEASES / P. Tirelli ; tutor: N. A. Borghese ; correlatore: N.A. Borghese ; direttore: E. Damiani. DIPARTIMENTO DI INFORMATICA, 2013 Feb 26. 24. ciclo, Anno Accademico 2011. [10.13130/tirelli-paolo_phd2013-02-26].
File in questo prodotto:
File Dimensione Formato  
phd_unimi_R08169.pdf

accesso aperto

Tipologia: Tesi di dottorato completa
Dimensione 3.22 MB
Formato Adobe PDF
3.22 MB Adobe PDF Visualizza/Apri
Pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/2434/217564
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact