Skip to main content

AUTONOMOUS CLOUD-COMPUTING VINEYARD ROBOT TO OPTIMISE YIELD MANAGEMENT AND WINE QUALITY

Final Report Summary - VINBOT (AUTONOMOUS CLOUD-COMPUTING VINEYARD ROBOT TO OPTIMISE YIELD MANAGEMENT AND WINE QUALITY)

Executive Summary:
Vinbot is an all-terrain autonomous mobile robot with a set of sensors capable of capturing and analysing vineyard images and 3D data by means of cloud computing applications, to determine the yield of vineyards and to share this information with the winegrowers. Vinbot responds to a need to boost the quality of European wines by implementing precision viticulture to estimate the yield (amount of fruit per square meter of vine area).

The Vinbot project aims to provide a solution to inaccurate visual sample-based surveys that are traditionally used to estimate vines yield and apply canopy management techniques accordingly. Using autonomous navigation technologies, a variety of sensors and advanced data processing techniques, the Vinbot can autonomously capture key indicators related to vine productivity and canopy features at exact positions throughout the entire vineyard and with minimum human intervention. A small onboard computer offloads data-intensive computer vision algorithms to be processed on the cloud. Using the web-based yield maps generated by the Vinbot system, winegrowers and associations can centralize and coordinate yield management throughout their members' vineyards.

Vinbot has been developed by a consortium of organizations including winegrower associations, technology and IT companies, agro service companies and research institutes, and has been co-funded by the European Union’s 7th Framework Programme. Working together since January 2014, the cooperation between these organizations has resulted in the achievement of the major objectives of the project. The project outcomes that have potential use either for commercial and research purposes are:
• Ground Truth database and correlation models
• Autonomous mobile platform with sensor head
• Robotic field navigation algorithms
• Mission planning web application
• Data management cloud software and web-based user interface
• Sensor data analysis, feature extraction and yield estimation algorithms

Vinbot addresses the strategic objectives of the European Commission’s wine sector reform through improved yield management by introducing a novel, autonomous, computer vision, and cloud-computing based robot economically accessible to SME winegrowers. In the field of the EU robotics sector, the diversification of the robotics sector is a European priority, moving away from manufacturing to new sectors. The general strategic behaviour of the major EU players is to seek new segments and expand out of the conventional industrial robots market.

The Vinbot consortium has been very active in performing dissemination activities. Main dissemination channels have been online channels such as the project website (www.vinbot.eu) each partners’ corporative website, Youtube and social networks, as well as traditional channels such as press, attendance to conferences and fairs and organization of events. During the 3-year project life Vinbot has been presented in 5 relevant industry fairs, 4 academic conferences, one publication, four papers, two posters, 5 articles published in popular press, 17 press release, 2 thesis and 1 workshop and in 2017 it is planned a report about the project to be featured in the Futuris TV program of Euronews channel.

In addition, a promotional video clip for public dissemination has been released to explain the goals of the project and show the results obtained (https://youtu.be/DZXmBPOiEfQ).
Project Context and Objectives:
Vinbot is an all-terrain autonomous mobile robot with a set of sensors capable of capturing and analysing vineyard images and 3D data by means of cloud computing applications, to determine canopy features and the yield of vineyards and to share this information with the winegrowers. Vinbot responds to a need to boost the quality of European wines by implementing precision viticulture to estimate the vegetative and reproductive components of a vineyard canopy. Winegrowers need to be able to estimate yield accurately in order to obtain robust information to help decisions on canopy management techniques, cluster thinning needs, planning and organize the harvest, planning cellar needs, planning purchases and/or grape sales, establish grape prices and management of wine stocks, etc. For a wine producer association, accurate yield predictions also allow to organize the reception at the winery, production and marketing of their members' wines. At the present moment most SME wine grower associations have little or none control over yield management throughout their members’ vineyards. Conventionally wine growers use inaccurate visual sample-based surveys to estimate yield which are very time consuming and, in most part of the cases, very inaccurate.

The Vinbot project aims to tackle these challenges through the development of an autonomous mobile robot enhanced with cloud computing applications to automatically acquire, process and present comprehensive and precise information on canopy features and grape yield to wine growers and associations in the form of a web-based map. Through their collective expertise in viticulture, the wine growers and associations can then coordinate crucial yield management techniques to improve efficiency and wine quality, per their commercial strategies.

Vinbot automates the traditional visual yield estimation process. Using a variety of sensors (detailed below) and computer vision, the Vinbot can autonomously capture the canopy features (e.g. exposed leaf area, canopy porosity) and grape yield at exact positions throughout the entire vineyard. A small onboard computer offloads data-intensive computer vision algorithms to be processed on external internet servers (a cloud software service). Using the web-based yield map generated by the Vinbot system, winegrowers and associations can centralize and coordinate yield management throughout their thousands of members' vineyards.

Each Vinbot can autonomously monitor hundreds of hectares several times a year. A trained, but not necessarily expert employee sets the robot’s mission settings, supervises it with very low degree of assistance and transports it from site to site. Due to the robust map-building and path-planning capabilities of the system the Vinbot is able to navigate intelligently and autonomously through the vineyard, collecting localized vegetative and reproductive data throughout the vineyard.

In concordance with the previously explained context of the precision viticulture, the general objectives of the Vinbot is to develop a solution that obtains realistic and accurate information on canopy characteristics and grape yield estimations from data obtained in the vineyards. This solution has five main elements, namely:
• the corpus data hand gathered by the experts participating in the project, which constitutes the ground truth database that was used to obtain correlation models and to test the output of the Vinbot algorithms.;
• an autonomous mobile robot system that allows the navigation in the field and sensor data capture;
• the sensor unit that encompasses a series of sensors to obtain relevant data to estimate canopy characteristics and yield;
• the processing software that handles such big amount of sensor data and transmits it to the cloud;
• and the cloud-based software that processes the sensor data, analyses images, creates heat maps and provides a visualization user interface.

The Vinbot offloads computer-intensive tasks like data and image processing and map building by processing the data gathered in the vineyard on the cloud.
Project Results:
Vinbot has been developed by a consortium of organizations including winegrower associations, technology and IT companies, agro service companies and research institutes, and has been co-funded by the European Union’s 7th Framework Programme. The project partners included the agricultural associations GRANJA (Portugal), PROVIR (Spain) and OGSZ (Hungary), the IT company ASSIST (Romania), the agro consulting firm Agri-Ciência (Portugal) and the winegrower Cantine d’Alfonso del Sordo (Italy). The R&D and demonstration activities were performed by the R&D company Ateknea Solutions (Spain), the Instituto Superior de Agronomia/Universidade de Lisboa (Portugal), and the mobile service robotics firm Robotnik (Spain).

Working together since February 2014, the cooperation between these organizations has resulted in the achievement of the major objectives of the project.

The following main scientific and technological results have been achieved, and are briefly described below.
• Ground Truth database and correlation models
• Autonomous mobile platform with sensor head
• Robotic unit navigation algorithms
• Mission planning web application
• Data management cloud software and web-based user interface
• Sensor data analysis, feature extraction and yield estimation algorithms
• Validation results

Ground Truth database and correlation models
The methodology for field data collection and generation of a Ground Truth (GT) database was defined at the beginning of the project consisting in the plans and protocols for data collection. Those plans encompassed the main following items: definition of the structure of the vineyard fields (vineyard plots, training system, varieties, etc.) and of the procedures to obtain the data (experimental design, type of measurements, methodologies, data collection calendar, etc.).
During the first two seasons (2014 & 2015), it were used only vineyard plots located at Instituto Superior de Agronomia (ISA) campus in Lisbon (2 vineyard plots in 2014 and 6 in 2015 encompassing white and red varieties). During the third season (2016), in order to provide a wider range of varieties and sites for validation purposes, the number of sites and varietal plots were expanded to 4 sites and 27 varietal plots (17 different grapevine varieties).
Each selected vineyard plot was labelled and submitted to several assessments per meter of canopy length in order to provide canopy, yield and vigour data. Furthermore, a set of detailed measurements per vine shoot and per cluster were also assessed on a sub-sample of vines per variety following several steps of a progressive defoliation and cluster thinning experiment. The process of data gathering was mainly concentrated between the phenological periods flowering to full ripening but some data was also collected during winter (pruning weight) and Spring (inflorescences).
The GT data collected was manually labelled and organized in a database format aiming at providing reference data for validation of results obtained automatically by Vinbot system. This database was also used for calibrating and adjusting the feature extraction and yield estimation algorithms.
Canopy characteristics (canopy dimensions, leaf area, shoot number, pruning weight) and yield components (crop load, bud fruitfulness, cluster number, cluster weight, number of berries per cluster, berry weight and yield) were organized per variety, per plant and per m of row length in order to allow the comparison with the output of the Vinbot sensors. The data collected was submitted to summary statistics and plotted in order to characterize and evaluate the vineyard vegetative and reproductive variability. Some of the obtained data was also submitted to correlation and regression analyses in order to study the relationships between variables and find explanatory variables to estimate vegetative and/or yield components. Regarding vegetative components, besides the expected variability between grapevine varieties, within the same vineyard plot it was observed a high variability in shoot number and length, primary and lateral leaf number and area and in canopy density and size. The correlation between vegetative variables showed that, with the exception of the primary leaf number, shoot length and number of leaves were significantly positively correlated with shoot leaf area indicating that those variables can be used as explanatory variables to predict shoot leaf area.
Concerning yield and yield components, it was also observed a big variability between grapevines varieties and that, within the same vineyard plot, vines presented a high variability in the number of clusters and yield. Detailed measurements also showed a high variability in cluster characteristics (weight, volume, number of berries and rachis length) among varieties and vines and between clusters within the same vine. Cluster number was well correlated with yield and, in most part of the varieties, cluster weight was significantly and positively correlated with the variables weight of berries per cluster, cluster volume, number of berries per cluster, rachis length and number of branches per rachis. The high and significant determination coefficients obtained indicates that most part of those variables can be used as explanatory variables to predict cluster number and weight.
Furthermore the data generated in the GT was also explored with the aim to obtain empirical models needed to complement the image analysis algorithms for yield estimation. Using a set of cluster data obtained in detailed measurements (cluster weight and corresponding projected area obtained in an RGB image made at the lab) very strong and significant relationships (R2 ranging from 0.78 to 0.96) were obtained enabling us to use the projected area of the cluster on the image as a good estimator of the corresponding weight. This model was obtained for each of the 8 varieties (4 white and 4 red) from the ISA vineyards and a general model was also obtained using pooled data from the 8 varieties.
In order to overcome the problem of non-visible clusters (hidden and/or partially occluded clusters) another empirical model was proposed in which the canopy porosity (obtained using the range finder) is used as an explanatory variable for the percentage of exposed clusters . To obtain those models it were used canopies with different levels of cluster zone defoliation (non-defoliated, half defoliation and full defoliated) from four varieties (2 red and 2 white). All obtained regression analysis showed a high and significant coefficient of determination (R2 from 0.18 to 0.74) however some of the relationships still showed a low percentage of variability explained indicating the need for further research into this modeling approach.
In parallel with GT data collection the selected vineyard rows were also scanned with Vinbot sensor head mounted on a trolley (2014 and 2015) or on the robot (final version of Vinbot platform, 2016). During the 3 seasons, a total of 272 missions were performed (80 in 2014; 80 in 2015 and 112 in 2016).
For the development of the computer vision algorithms the cluster projected area on the RGB images was labelled manually (using squares for delimitating the area of visible berries) on more than 7000 images (1/m canopy length) collected on both sides of the canopy in a wide range of canopy densities (high, medium and low density canopies, and on full defoliated canopies).

Autonomous mobile platform with sensor head
A modular robotic unit consisting in a mobile platform and a sensor head has been designed and build. The unit is made of commercial off-the-shelf components. The mobile platform is based on the Summit XL platform by Robotnik and the sensor payload has been selected through extensive testing in order to find a compromise between cost and efficiency in both navigation and vineyard data acquisition. Sensor payload include a 2D laser rangefinder and a Kinect v2 camera for capturing vine data and a 2D laser rangefinder, an inertial measurement unit and a RTK-DGPS localization unit for robot navigation functionalities. Vine data is acquired as the robot moves forward and synchronized with the navigation sensors (Fig. 1). The complete robot software architecture, including the software for capturing viticulture data, has been developed in ROS (Robot Operating System), an open source and free software framework for robot software development.
Two prototyping iterations along with intensive laboratory and field testing has allowed to reach a fully functional unit which meets mobility and data acquisition requirements in most common scenarios.

Robotic field navigation algorithms
The robotic unit can autonomously navigate over semi-structured environments such as vineyards thanks to a set of algorithms that have been developed in the project. The navigation is based on a hybrid reactive navigation scheme that requires just a 2D laser scanner and an accurate DGPS localization and combines reactive navigation and fixed waypoint routes previously defined as a set of parametrized moving instructions (e.g. “follow row center”, “turn right”).

Mission planning and robot monitoring web tool
The Vinbot robot integrates a mission planner that allows the user to easily set up a mission through a web interface. A minimum set of characteristics of the vineyard are requested such as the distance between rows. The mission is defined by a sequence of waypoints that are parametrized by reactive (waypoint type, direction and speed) and location-based parameters (move to relative location or move to absolute location).
The user can supervise the execution of the mission by the robot as well as monitor the current status of the mobile unit by means of a specific web interface (Fig. 2)
Both tools are accessible from any standard web browser on a PC or handheld device that is wired or wirelessly connected to the robot unit either through WiFi/IP-Radio or mobile network (depending on the wireless communication option mounted on the unit).

Data management cloud software and web-based user interface
All the data collected by Vinbot sensors during a mission including both vineyard and navigation data is packed in a specific format and transmitted at the end of a mission to a remote server where the Vinbot cloud software is running. This software has been developed to manage and process the data files generated by the robot and extract useful information from them. The size of data files as well as the compute-intensive algorithms required to perform particular analyses such as image processing justify the use of a cloud infrastructure, which drastically reduce processing time and computation cost compared to on-board processing solution.
The data management software consists of a set of backend software components that implement a workflow specifically designed to manage incoming Vinbot sensor data files, execute diverse data analysis routines in an ordered manner and store the results.
Furthermore, this software includes a data visualization tool to allow end users access to the collected data and visualize the calculated indicators by means of heat maps projected over satellite images of the vineyard from Google Maps (Fig. 3). The data visualization tool again consists of a set of backend components dealing with data access/data exchange, security, user management, session management, and map building functionalities as well as of a web-based graphical user interface.

Sensor data analysis, feature extraction and yield estimation algorithms
The Vinbot cloud software features a sensor data analysis software module that is in charge of obtaining valuable insights from raw sensor data. This is done by merging wisely and processing GPS data, odometry data, RGB pictures and 2D rangefinder data to obtain key features of every segment of the vineyard. Later on, the correlation models are applied on some of the indicators in order to obtain the yield estimations.
Two types of indicators are obtained, yield indicators and canopy features indicators. Yield indicators are calculated by processing the RGB pictures with a specially-trained deep learning model which is capable of recognising and quantifying entire or fractions of grape clusters either exposed or partially occluded by leaves under a wide range of light conditions. Canopy features indicators such as canopy porosity index, canopy height or canopy volume are obtained by analysing the 3D shape of the vine that is generated by combining the 2D rangefinder and robot’s forward movement. This analysis is automated by means of a set of dedicated and tailored algorithms.
The module generates an output file in XML/CSV format containing the calculated indicators for every vineyard segment along with its geographical position. This makes this file easily exportable to third party software tools such as GIS tools for its use in other purposes.

Validation results
In order to test the behavior and robustness of the complete Vinbot system in real field conditions a validation plan was developed in WP7. This validation plan was used to guide the tests and demonstration of the Vinbot system in vineyards of some consortium partners during the second half of the 2016 growing season. The validation plan has taken into account the functioning and performance of the autonomous platform and the manual assessment of vegetative and reproductive parameters to compare with the output of the Vinbot system.
The field validation (Task 7.2) had as main objective to provide a clear insight into the real robustness of the system, as well as to pinpoint which features need to be fine-tuned and how to proceed in order to easily adapt the Vinbot platform in a standardized way. The field validation was performed in Portugal during the summer 2016. A total of 112 sessions were accomplished in different Portuguese vineyard conditions (different sites, varieties, soil and canopy management, slopes, etc). Besides the manual assessment of vegetative and reproductive variables and the image and range finder data collection, it was also tested the robot performance (autonomy, recognition of the end of the row and turn; sensitivity of the navigation sensor, etc.) and trained the users in using the smartphone application.
In what concerns the functioning and performance of the autonomous platform, despite some minor problems, in general the robot was able to deal with most field and canopy challenges. Most part of the navigation problems were solved and others have aroused suggestions for future improvements. The worst problems were found in steep slope vineyards with recently tilled soils. In those cases, the robot showed difficulties regarding its locomotion, due to skidding, which resulted in repetition of image recording every time the robot was not moving forward.
Regarding the comparison of actual and estimated single values (per m of canopy length) of canopy height the best fit was observed in the vineyard plots of Alvarinho and Viosinho (ISA), Tinta Caiada (Granja), Tinta Miúda (Quinta do Pinto) and Chardonnay and Touriga Nacional (Quinta da Amieira) were data showed a satisfactory agreement and an acceptable error. In all the remaining vineyard plots a low or absence of agreement was observed (over and underestimation) with an error above the upper limit of acceptability. The validation results for the two other canopy features (exposed leaf area and canopy volume) showed a similar fit to that reported above for canopy height. The cases of low agreement between the individual values of canopy features per m of canopy can be explained by several reasons being the possible desynchronization between ground truth data and corresponding Vinbot images one of the most important. Indeed when using averaged data per smart point (5-10 contiguous meters of canopy length) the agreement between actual and estimated values has improved and a strong and significant linear relationship between actual and estimated values combined with a lower error was observed.
Concerning the yield estimation, the validation results showed a low fit between single values of actual and estimated yield per m of canopy length. Those results are very disappointing as yield estimation was the core of Vinbot project. Indeed, in general (except for the variety Touriga Nacional at the site Quinta da Amieira) a low or absent agreement and a very high error was observed between actual and estimated single yield values. The Vinbot algorithms underestimated the yield for ISA plots and overestimated for Granja, Quinta do Pinto and Quinta da Amieira ones (except for Chardonnay, Aragonez and Touriga Nacional) with an error above the upper limit of acceptability. This low agreement between actual and estimated single yield values per m of canopy length can also be explained by the possible desynchronization between ground truth and corresponding Vinbot images. Indeed very often the Vinbot images included more than 1 vine (e.g. half a meter of two contiguous vines). However, when using pooled data (average values of 5-10 m per smart point) despite the significant R2 showed by the regression analysis between actual and estimated values, the percentage of variability explained is still low and the slope of the fitted line indicates that the Vinbot algorithms still overestimate the yield by a multiplicative factor. This low accuracy enables us to conclude that further research on computer vision algorithms, modeling and data processing is needed to improve the Vinbot prediction ability.
Potential Impact:
Potential impact
Although its market leader position, the EU wine sector suffers from serious structural shortcomings, the most relevant being the inflated surplus of wine. The survival of European SME vineyards is threatened by current market forces. As most European vineyards are relatively small, the large majority are part of associations like the Vinbot SME-AGs. The associations centralize and optimize production, as well as distribute and market their members' wine. Cost-effective, centralized, automated yield management is a critical tool to boost quality, productivity, competitiveness and brand recognition in the European wine sector and allow EU wine producers to embrace precision viticulture technologies.

The Vinbot addresses the strategic objectives of the European Commission’s wine sector reform through improved yield management by introducing a novel, autonomous, computer vision, and cloud-computing based robot economically accessible to SME vineyards.

Another advantage of Vinbot is that it is an open robotics platform, comprised of off-the-shelf sensors. As such, other sensors may be added, and the web services can be configured according to the needs of the industry to measure additional vineyard conditions in the future. The proposed technology can work with existing standard sensors to monitor temperature, humidity, diseases, vegetative stress, as well as other standardized sensors developed in the future.

In the field of the EU robotics sector, the diversification of the robotics sector is an European priority, moving away from manufacturing to new sectors. The general strategic behaviour of the major EU players is to seek new segments and expand out of the conventional industrial robots market.

There will also be an impact in the standards of the future of agriculture and viticulture that will rely in precision agricultural tools, specifically robotics. The ability to autonomously carry out specific actions, whether they be crop monitoring, planting, harvesting, etc., has huge implications in the agricultural sector, where traditional agricultural methods are not enough to sustain the competitiveness of European agriculture. Complementary emerging technologies, such as agricultural robotics, are increasingly accepted by farmers, due to their cost effectiveness, and the automation of work that is dirty, difficult and dangerous.
Environmentally, the Vinbot system uses electric energy (lithium-ion batteries) to power the mobile robotic platform with no CO2 pollution. In a broader point of view, the Vinbot will compile a database of visual vineyard characteristics online. Although this is beyond the scope of this current project, future research could be carried out regarding the effects of climate change and other environmental considerations at the vineyard through the use of this database (always with the explicit authorization of the parties involved). A key aspect of winegrowing is that is essential to avoid desertification, which is an on-going source of environmental concern, especially in the Mediterranean basin.

A common fear with respect to robots is that they create unemployment. This is quite on the contrary and it is being widely stablished that the industries where robots are more present tend to generate more and better jobs. In particular, we think this could also happen in rural economies, which have particularly suffered the hardest by the global financial crisis. Robots deployed on the field will be accompanied by non-expert personnel to set up the basic functions of the system and area of work, and monitor its proper functioning. Besides, new consulting services will emerge that will use Vinbot as a tool for obtaining information and drive a transformation towards precision viticulture and make them set up new business models. In this respect, the Vinbot will create direct employment opportunities. Furthermore, through increased revenue generated by use of the system, the successful implementation of the Vinbot will encourage indirect employment by injecting revenue into struggling rural economies. Finally, in line with the EC’s objectives to create the world’s most competitive knowledge based economy, the Vinbot project will boost employment and the international profile of the EU robotics sector by creating the first agricultural cloud-computing robot.

Main dissemination activities

The Vinbot consortium has been very active in performing dissemination activities. Main dissemination channels have been online channels such as the project website, each partners’ corporative website, Youtube and social networks, as well as traditional channels such as press, attendance to conferences and fairs and organization of events.

A website of the project has been created at www.vinbot.eu to keep a wide general audience updated about the progress of the project and to provide contact information of the members of the consortium. In addition, a promotional video clip for public dissemination has been released to explain the goals of the project and show the results obtained. This video aims to explain in an understandable way how the Vinbot system works and why it can help winegrowers in keeping a better control of yield and quality of their vineyards. The video clip is available on the news page of the project website (http://vinbot.eu/news) and on YouTube (https://youtu.be/DZXmBPOiEfQ).

Other relevant dissemination actions carried out during the project are listed below:

Publications:
• R. Guzman, R. Navarro, M. Beneto, D. Carbonell, “Robotnik - Professional Service Robotics Applications with ROS.” - Chapter of the Springer ROS Book – The Complete Reference (Volume 1), P253-288. Ed.: A. Koubaa. ISBN 978-3-319-26054-9. DOI 10.1007/978-3-319-26054-9. The chapter includes an important introduction to the Vinbot project as an autonomous agriculture mobile robot. January 2016.
Conference presentations/papers/posters:
• Braga R., Graça J., Lopes C.M. 2015. “Viticultura de precisão – Casos de Estudo em Portugal” Simpósio Vitivinícola da Regiões de Lisboa, Tejo e Península de Setúbal, 19 e 20 de Novembro, 2015, Almeirim, Portugal (Book of Abstracts).
• Lopes, CM; Graça, J.; Sastre, J.;.Reyes, M.; Bautista, J.; Guzman, R.; Braga, R.; Monteiro, APinto, P.A. 2016. "Estimativa automática da produção de uvas utilizando o robô VINBOT – Resultados preliminares com a casta Viosinho", 10th Simpósio de Vitivinicultura do Alentejo, May 4-6, 2016, Évora, Portugal. (Oral presentation)
• Lopes, CM; Graça, J.; Sastre, J.;.Reyes, M.; Bautista, J.; Guzman, R.; Braga, R.; Monteiro, APinto, P.A. 2016. “Vineyard yield estimation by Vinbot robot – preliminary results with the red variety Trincadeira” X International Symposium on Grapevine Physiology and Biotechnology, 13-18 June,2016 , Verona, Italy (Book of Abstracts, 234); (poster presentation)
• Lopes, CM; Graça, J.; Sastre, J.;.Reyes, M.; Bautista, J.; Guzman, R.; Braga, R.; Monteiro, APinto, P.A. 2016. “Vineyard yield estimation by Vinbot robot – preliminary results with the white variety Viosinho”. XI International Terroir Congress, Willamette, Oregon, July 10-14, 2016 (Poster presentation)
• Guzmán, R. 2016. “Autonomous hybrid gps/reactive navigation of an unmanned ground vehicle for precision viticulture - VINBOT” Technical presentation and paper at Intervitis Interfructa Hortitechnica - Technology for wine, juice and special crops, Stuttgart, Germany, November 27-30.
To be submitted in the next months:
• Lopes, C.M.; Graça, J.; Reyes, M.; Torres, A.; Guzman, R.; Vitorino, G.; Braga, R.; Monteiro, A.; Barriguinha, A. 2017.“Using an unmanned ground vehicle to scout vineyards for non-intrusive estimation of the canopy features and grape yield”, 20th Groupe International d’Experts en Systemes Vitivinicoles pour la Coopération, Mendoza, Argentina, November 5-9, 2017.

Other dissemination actions:
• The COST Action – Quality Fruit was realized in ISAat the 29th September, 2015. During the event, the project Vinbot was presented to several international researchers (ca 30) and a field demonstration was done.
• Organization of the “1st Workshop on Smart Viticulture”, presentation and field demonstration of Vinbot, Évora, Portugal, April 29, 2016.
• Presentation of Vinbot (including the robot prototype) in AgroGlobal, an important professional agricultural fair in Portugal, in the 2016 edition that took place in Valada do Ribatejo on 7-9 September 2016. 250+ companies, 20,000+ visitors.
• Presentation of Vinbot project at the Annual General meeting of “Lien de la Vigne”, within the topic “Nouveaux outils pour le suivi de la qualité des raisins: capteurs, analyse des données, outils d’aide à la decision”. Paris, France. March 31, 2017.
• Vinbot project featured in Futuris program in Euronews. Field demonstration to be prepared and filmed in June or July 2017, along with interviews to consortium members.

Exploitation of results

From a commercial point of view it can be stated that the Vinbot system as integrated solution is at level 7 of TRL (Technology Readiness Level) thus not ready for commercial launch, but that the project is aiming at the good direction. Some of it components such as the mobile platform and navigation features are even closer to market. However, it is clear to the Consortium that further research on computer vision algorithms, sensor data processing and modelling is needed to improve the Vinbot prediction ability, as well as more refinement of specific on-field robot operations and robot handling. Despite this, the Consortium envisages a clear business case with relevant market potential and has defined a development roadmap and a commercialization strategy in view of future market launch and provided that funding is raised to finance the remaining work.
List of Websites:
www.vinbot.eu

Contact e-mail address: info@vinbot.eu andrebarriguinha@agriciencia.com info@agriciencia.com