Non-Invasive assessment of fruit – improving quantity and quality
Published:13 July 2018
A research team, led by CQUniversity’s Professor Kerry Walsh (pictured TOP left), has been working in collaboration with fruit growers, horticultural industry bodies and technology solution providers to develop technologies that can help farmers improve their yields and potentially automate on‑farm processes in the future.
Northern Australia is famous for its sub-tropical climate. This climate allows the region to produce some of the world’s best and freshest tropical fruits for local sale and export. But, there are disadvantages – the region is far from large population-centre markets, and at the end of supply chains for labour and materials.
Growers are constantly trying to find better ways to plan crops and resources, increase production, market their product, manage their supply chain and achieve better financial returns. This need has led the industry to investigate how emerging technologies may be able to help growers improve yields and on-farm efficiencies.
A research team, led by CQUniversity’s Professor Kerry Walsh, has been working in collaboration with fruit growers, horticultural industry bodies and technology-solution providers to develop technologies that can help farmers improve their yields and potentially automate on‑farm processes in the future.
This research direction was instigated by an approach to researchers from Central Queensland pineapple growers, almost two decades ago, with the aim of finding a way to prove that the region's produce was sweeter than that of their South East Queensland counterparts.
Near Infrared Spectroscopy (NIRS) technology was identified and developed for in-line (packhouse) sorting, in conjunction with MAF Oceania Pty Ltd, and handheld (in-orchard) assessment, in conjunction with Felix Instruments Inc. Pineapples proved to be a difficult subject, due to their rough, thick skin. However, it was found that the technology suited thin-skinned fruit, such as apples and mangoes. It was found this technology was useful in providing an objective measure of fruit on the tree, guiding the decision on when to harvest. This type of information is invaluable to producers as harvesting too early results in immature fruit reaching market, while harvesting too late results in fruit that are softening and do not pack and transport well.
Now, two decades later, the research and technology has evolved into a multi-pronged approach to assess fruit quantity and quality of fruit on tree, in the orchard, and to assist farm management decisions. Current projects include in-field machine vision for estimation of tree flowering time, and estimation of fruit number and size, and a wireless orchard temperature logging system. These measures can assist in identifying the optimal time to schedule the harvest, amount of labour required for the harvest, and resources and logistics to get the fruit to market.
Data, however, is just numbers. That is why the research group has been developing a web application to turn complex data into information and knowledge. The app presents the various data streams visually, allowing for ease of use by growers. Combined with the growers’ experience and knowledge of their crops and farm management, growers can use the web app to strategically manage their crops and harvest resources, creating better efficiencies and reducing waste.
This research and development work has been funded over an extended period by Hort Innovation Australia (HIA), with the research team working in close collaboration with the Australian Mango Industry Association and specific growers.
In recent years, the Australian Government Department of Agriculture and Water Resources, as part of its Rural R&D Profit program, has also supported the work through HIA, in context of a larger project (involving UNE, Uni Sydney etc). In the CQUni component of this work, machine vision using cameras on farm machinery in fruit orchards has been used to record and map the extent of tree flowering. Early flowering trees will have early maturing fruit, which can be selectively harvested and targeted to particular markets.
With the start of flowering mapped, the time to fruit maturity can be estimated from so-called ‘heat sums’ that integrate temperature across time. However, collecting temperature data on farm is another tedious task for the grower, so the team are developing a system based on low-cost wireless temperature recorders, placed throughout orchards, that record and transmit temperature conditions to the app, with automatic updating of the heat sum and projection of likely maturity/harvest date.
The machine vision cameras used to record and map the extent of tree flowering are also being trialled in estimation of fruit number on the trees. This activity can help to provide growers with an early assessment of crop quantity, allowing them to plan harvesting and packing resources and estimate possible crop returns.
Finally, with crops near maturity, the handheld NIRS instrument can be used to assess on-tree fruit quality in the orchard. These sensors determine the fruit’s dry matter content, which is an index of fruit starch and sugar content. Essentially, this index correlates to how good the fruit will be, come time to eat. A low dry matter content means fruit is not yet ready to harvest, while a higher reading means fruit is reaching the right level of maturity for harvest. This means growers can harvest fruit at optimal times to improve the ease of harvest and packing and ensure the eating quality of a crop. The instruments have GPS capability, so all data can be easily shown on the app, with automatic calculation of block averages and rate of increase.
The CQUniversity research team is currently working primarily with three fruit growers as part of this project – mango and avocado farms near Childers and Rockhampton, and a mango farm near Darwin.
Acacia Hills Calypso Mango farmer Martina Matzner has worked with the team for five years to plan harvests and manage resources. She said that the technology had provided invaluable information about getting the timing of harvest right, leading to improved pack out, recruitment of harvest and sources and supply chain management.
‘The use of this technology has allowed us to more precisely plan our labour for harvest and importantly allows us to plan the order of harvests for blocks based on what trees are bearing the most mature fruit.
‘It means we can improve efficiencies, reduce waste and increase returns,’ said Ms Matzner.
Working with technology solution providers MAF Oceania and Felix Instruments, as well as several industry partners, the CQUniversity research has been commercialised, with research and development continuing to expand application to other fruit crops, such as avocado.
Professor Kerry Walsh said that as well as expanding the research to other fruit, the group would continue to refine the technologies in issue.
‘It is an exciting period, with amazing developments occurring in technology such as NIRS sensors, machine vision and autonomous vehicles. All these developments can be put to use when it comes to a variety of Australian tree crops.
‘The long-term objective here is automated harvest. Now we have machine vision ‘seeing’ the fruit for counting purposes – it's foreseeable that automated harvest is possible.
‘Mango crops are a natural fit for this. The fruit is large and relatively robust, and finding harvesting labour willing to work in 40 degree Celsius heat and endure the acidic sap burns that come with harvesting the fruit, is often difficult for growers,’ said Professor Walsh.
The research team will continue to trial the technologies and methods for automated harvest with mango growers and is also exploring opportunities to use the technology in avocado and stone fruit crops.