Month: November 2013

Semi-automated and automated assessment of sparkling wine quality based on foam stability, assessed using a robotic pourer and image analysis algorithms

By Sigfredo Fuentes; Bruna Lima, Maeva Caron and Kate Howell


Figure 1: 3D model of the robotic pourer.


Figure 2: Inside and outside of actual pourer.

The Australian wine industry contributes strongly to the country’s economy (Lin, 2013). Australia is the sixth largest wine producer and the fourth largest wine exporter (WFA, 2013). Sparkling wine accounts for approximately 9% of the domestic wine sales and 13% of total wine imports in Australia (ABS, 2013). Wines with dissolved carbon dioxide are also economically important for several countries, France, Spain, Italy, USA, and Chile (IWSR, 2011; CIVC, 2013). Increase in temperature and carbon dioxide levels in the atmosphere have shown to affect flavour and aroma of wines (Johnson and Robinson, 2013), decrease protein concentration in plants (Högy et al., 2009; Azam et al., 2013; Mishra et al., 2013) and increase alcohol in wines (Howden and Stokes, 2009). Consequently, climate change is expected to influence the final quality of sparkling wine, since several compounds, including protein concentration and alcohol are related to foam stability and the ability of the wine to produce foam (Pozo-Bayón et al., 2009; Coelho et al., 2011).

Figure 3: Sparkling wine pouring

The quality of sparkling wine is visually assessed by its colour, bubble behaviour, appearance (bead) and foam persistence (mousse) (Liger-Belair, 2013). However, as discussed by a number of authors, these parameters are extremely variable and are affected by pouring, reception vessel shape and type as well as temperature (Cilindre et al., 2010; Liger-Belair et al., 2012; Liger-Belair et al., 2013). Robotics and chemometrics allows us to control and monitor these parameters, and thus repeatedly measure sparkling wine for quality assessment and to correlate it with traditional measures of quality. A robotic bottle pourer has been developed to standardise time and wine volume of pouring into a standardised vessel. Images are collected automatically with a digital camera attached to the pourer and transferred to a computer. These images are then evaluated by image analysis algorithms, which convey the information into bubble size and speed, foamability (ability of the wine to produce foam), foam persistence and stability, and collar stability.


Figure 4: Pourer in action in parallel to a sensory analysis at Moet Chandon, Yarra Valley, VIC – Australia.

As a result, this robotic pourer and image analysis algorithms, which simultaneously quantify both bubble’s individual behaviour and collectively as part of the foam, allows the development of a reproducible, easy and inexpensive method to measure sparkling wine quality. Results from this novel technology have been compared to chemometrics and sensory panel data using multivariate data analysis.

The use of multicopters for vineyard monitoring

By Sigfredo Fuentes


Figure: Octocopter from the VoF – Chile carrying an infrared thermal camera, visible camera and multispectral camera.

The MSLE with the Department of Mechanical Engineering from The University of Melbourne are developing customised unmanned aerial vehicles (UAVs) in the form of multicopters. The main aim of this project is to develop continuous remote monitoring systems to assess spatial and temporal variability of plant growth and water status, combined with a decision-making tool based on processed GIS maps (GIS-DMT). These tools will enable grapegrowers to apply efficient management strategies to maximise genetic cultivar potentials. Currently available methods to assess plant vigour and water status rely on expensive portable or fixed location instrumentation making the assessment of spatial and temporal variability of these parameters within a grapevine field difficult. Spatial variability can be obtained from airborne sensors and satellite imagery, which can be cost prohibitive and also have operational complexity in the information handling and data interpretation. Satellites also have a spatial resolution problem and long revisiting periods (low temporal resolution). Advances in miniaturised remote sensing technology (visual, infrared thermography and multispectral cameras) allow mounting them on UAVs and UTVs to assess plant vigour and water status using visible, multispectral and infrared thermal images. These systems are of high temporal and spatial resolution, which can be combined with GIS-DMT for rapid automated data analysis and reporting. The expected outcome would be management GIS maps of grapevine fields accessible to grapegrowers through a web page and specialised smartphone/tablet PC App. This integrated tool will offer competitive advantages to the Australian wine industry at a reduced cost compared to other available technologies.

The use of Dogs in Viticulture / Oenology

By Sonja Needs and Sigfredo Fuentes (The University of Melbourne)


Dog in the picture: Luther, trained by Sonja Needs.

The use of dogs for detection is nothing new. Dogs are used in a range of different detection endeavours including detecting drugs, explosives, ores and minerals, insect pests and cancer to name a few (Yinon, 1999; Furton and Myers, 2001; Bhadra, 2011). A dog’s olfactory cortex is 40 times larger than a humans and enables them to detect concentrations nearly 100 million times lower than humans. Dogs are able to recall isolated scents and can “store” scent in a nasal pocket created by a bony subethmoidal shelf. This nasal pocket permits odour molecules that are unrecognizable in a single sniff to accumulate and interact with olfactory receptors.

Early detection of pests and diseases in grapevines is vital to enable more effective control or eradication before symptoms occur or disease spreads. This project aims to train dogs to seek out and indicate the presence of three target odours; Brettanomyces, Eutypa lata and Phylloxera (Daktulosphaira vitifoliae). Determining firstly the ability of the dogs to detect the odour and secondly to assess reliability and thresholds for detection of that target odour. Six dogs of various breeds are currently being used for this project. Once trained, a dog can be switched from one scent to another relatively quickly. There are a large number of possibilities for use in the winery and vineyard. A properly trained dog can be more reliable than many current forms of detection, however one of the main reasons for this project is to ascertain the presence of odours for the potential for use of an electronic nose & chemosensors.