Technology prevents kiwifruit from going pear-shaped

Locally built hardware and software is helping the fruit industry sort the wheat out from the chaff

Making sure high quality fruit is actually high quality is quite an issue for the fruit industry not only here but all around the world. When you're working with hundreds of thousands of fruit per day, it can quickly become a nightmare.

Enter Compac Sorting Equipment, based in Onehunga in Auckland, which started as a small family company in 1984. Today, the company has seven offices and 160 staff world wide; a research and development centre in Auckland and manufacturing facilities in New Zealand, Korea and Italy.

The company spends about 10% of its turnover on R&D, says Nigel Beach, R&D manager of Compac Sorting Equipment. The Onehunga R&D centre has 12 software developers and another ten R&D staff, he says. All of Compac’s software is developed locally.

“Most of our R&D is in Auckland, but a little bit is done in Italy and Korea, where we also have factories,” he says. Staff with design skills or staff with a mathematical background are particularly hard to find locally, so Compac hires from all around the world. Half of Compac’s software is built using C++ and half on the .net platform using C#, says Beach. The developer team is using the agile development methodology extreme programming, he says, which focuses on adapting to changing requirements during a project, rather than trying to define and control all requirements at the beginning of a project. Compac also designs and develops its own electronics hardware, says Beach. One of the features that makes the software unique is a kiwi fruit blemish sorter.

The system, called InVision 9000, sorts citrus, kiwifruit and apples for surface marks, stains, insect damage, cuts, punctures and bruising with the help of cameras. There are four cameras for every lane of a machine, explains Beach. “A typical New Zealand customer might have four or eight lanes [whereas] the big systems in the US, for example, will have up to 40 lanes,” he says. The cameras take colour and infrared pictures of the entire fruit surface — 60 images per fruit, 10 fruit a second, he says. “600 fruit images per second are being analysed. [The system] works out the dimensions — works out if there are any defects in the size of the fruit. It looks at the colour of the fruit and looks for little marks on the fruit,” he says. The system analyses each mark, its size and position, while ignoring the normal marks, such as the ends of a kiwifruit, and works out whether it is a good fruit or not, he says. The main challenge when building software for these kinds of systems is dealing with a natural product — and a lot of it, says Beach. “An error of a few percentages could be thousands of fruit,” he says. “[Our customers] might be processing 500,000 to a million fruit a day.” In addition, it is quite difficult to do testing in the R&D centre because things like seasonal changes and sheer volume must be taken into consideration. “It’s an ever changing product that comes in differently depending on the season,” he says. “Vision systems that have been around before [have generally been for] manufactured products that look the same, whereas we are looking at whether something fits into quite wide criteria.” Compac has sold around 500 machines across the world, of which around 100 in New Zealand. The company has implemented one of the largest sorting lines in the world for Sun Pacific in California, says Beach. “[The line] is half a rugby field in size and is capable of [sorting] 500 mandarins a second. It’s got 40 computers to run the blemish system,” he says. In addition to the 40 lane machine, Compac has also implemented two 12 lane packing lines and another 20 lane sorting machine for Sun Pacific in California. Beach says that there is a lot more to sorting fruit than he could ever imagine when he started working for Compac a decade ago. Not only is fruit sorted for size, colour and blemish, it also sorted for sweetness and internal defects. Using NIR (near infrared) technology, the system determines the sweetness of a fruit by shining a light on it and measuring which wave lengths are absorbed and which are reflected back, says Beach. The fruit reflection is measured by a spectrometer — an extremely sensitive colour detector. So, without touching the fruit, the system can sort ten fruit a second for sweetness, he says.

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

Tags R&Dinfrared

Show Comments