As part of a larger R&D-programme, Wageningen University & Research is developing a toolbox that companies can use to check the quality of vegetables, fruit, flowers and potatoes at the start of fresh-produce chains. This will allow them to predict the quality of the products as they progress along the chain. The toolbox will make it easier to make the right decisions about logistics in the journey from greenhouse to kitchen.
At present, it is almost impossible to measure, quantify and predict the progression of quality. Chain condition monitoring and quality loss models are commonly used in the fresh-produce sector, but there is rarely a reliable, objective indication of the initial quality. The quality at the start of the fresh-produce chain determines both the progression of quality in the chain and the residual quality or shelf-life. The choice of variety is a determining factor for initial quality, as are the cultivation conditions and crop management, post-harvest treatment and storage history of the product.
The R&D-programme Quality Phenomics is developing knowledge and methods to help improve the quality of perishable products, safeguard quality aspects in the chain, reduce wastage from the chain and develop varieties that are better suited to conditions within the chain. The knowledge and methods used include robotics, internal structure analyses, spectroscopy, metabolomics and modelling. The programme is a partnership between experts from Wageningen UR in the fields of plant growing and cultivation, post-harvest physiology and advanced storage techniques, metabolomics, robotics, machine learning, sensors and imaging, and modelling techniques.
One of the tools being developed in the R&D-programme is a robot that can measure the quality of individual fruit and vegetable products on the basis of shape, colour and NearInfraRed (NIR) fingerprint. The results of the measurement are then translated into product quality readings via computer vision and machine learning techniques. Buyers and sellers of fruit and vegetables can perform a fully automatic entry check on their fruit and vegetables using the robot, without the risk of damaging the products. The robot will also help fruit and vegetable growers to screen and assess the quality of new varieties automatically. In both situations, the quality control is carried out visually and by analysing samples.
Wageningen UR has already constructed a demo model and work on the robot will continue for the next few years. The robot hand will be fitted with sensors, for example, to measure firmness, Brix and other characteristics used to assess quality. The focus will be on translating time-consuming human assessments and destructive measurement methods into non-destructive, objective measurements, which the robot can carry out automatically in batches per crate.
The research programme makes use of quality models to determine whether apples and pears from different orchards are suitable for long-term storage or export to faraway destinations. The models make it possible to predict quality and determine storage behaviour. The programme is also exploring whether the ripening process of tropical (and subtropical) fruit, such as mangos, avocados and bananas, can be predicted by measuring product characteristics when the product arrives in the Netherlands. The aim is to manage the ripening process more efficiently, reduce wastage and improve the quality of the fruit on the shelves. The initial firmness and size of the fruit at the start of the chain appear to be important factors for predicting the speed at which it will ripen. Acoustics is a good, non-destructive way of measuring firmness. Ripening can be managed more efficiently by classifying the fruit in terms of firmness.
In addition to the quality indicators already used, such as colour, Brix and firmness, the programme is also exploring potential new indicators. For example, by analysing the aroma compound profile with a PTR-TOF-MS or measuring structure properties using X-ray Computer Tomography (XRT), which compiles an internal structure analysis via 3D X-ray images. Measurements of this kind could possibly be future additions to the quality phenomics toolbox.