Dual Laser Technology

A rapid method using laser technology to non-destructively detect internal changes that may lead to early spoilage in fruit and vegetables has been developed by Plant & Food Research and is now being incorporated by technology company Edison Darby into a commercial scale in-line scanner for use in local and export industries.

Fruit and vegetable exports earn NZ more than $4 billion annually. In international markets, the presence of rot or spoilage can result in entire shipments being returned or destroyed, so being able to detect subtle changes in the flesh composition that may lead to premature spoilage of produce would allow exporters to remove potentially damaged items prior to packing and transportation. It can also allow intact produce with a shorter predicted storage life to be diverted to local or less demanding markets.

For many years, Plant & Food Research scientists at Ruakura have been studying the uses of near infra-red spectroscopy, for the non-destructive measurement of fruit properties. Dr Andrew McGlone, Principal Scientist & Science Team Leader for this group, says that a large component of this research has focused on the storage life of fruit. “The dual laser approach is a subset of technology that’s trying to detect small defects that are a disturbance to a consumer when they, say, bite into an apple and find that it’s rotten or brown,” he says.

“This project started in 2015 with Jason Sun's PhD, which he did with us, to optically measure structural and compositional information inside fruit and vegetables as a general topic, and we ended up with this dual laser system.” The new technology uses laser beams to scan produce as it moves through the packhouse grading system, identifying produce that might be defective or have incipient changes.

So far, the system has been shown to detect internal damage to apples, kiwifruit and onions, and with higher accuracy than existing scanning technologies. It involves a new type of near-infrared technology, termed Dual-Laser, that uses a pair of laser light sources modulated at different frequencies, and a pair of simple photo-diode detectors.

Using a fast rotating multi-sided, multi-angled mirror, the laser beams scan across the top half of the produce visible to the lasers, with the detectors collecting the transmitted light from the apple sides. The collected laser beam signals result in high resolution data from thousands of scanning locations being fed into a computer algorithm that has been ‘taught’ through machine learning to identify damage within the produce. Produce with damage can then be redirected, based on the extent of the damage.

Early trials showed that it was potentially a valuable addition to existing sensing technology and was picked up by a commercial company wanting to integrate the dual laser approach into a high-speed grading line. Unfortunately, the company involved got into financial difficulties.

Subsequently a Waikato based company has taken on the challenge. Edison Darby describes itself as an active venture studio, specialising in growing internationally capable, technology enabled businesses. Although it is a new company its founder, Geoff Furniss, and many of its staff have had long experience developing fruit handling and grading equipment at BBC Technologies.

“We are creating and developing businesses as standalone entities, one of which involves the commercialization of the Dual Laser system with Plant & Food Research. We've spent a lot of time on inspection systems for the produce industry, so we’ve got a very solid understanding of most of the technical methods for identification and classification of produce.”

Geoff says they were looking at novel sensing methods and realised the dual laser approach could be a useful addition. The company has already done a lot of work using infrared light to inspect surfaces. Adding a dual laser scanner has allowed examination of internal tissues but has required a lot of testing to find specific wavelengths that identify specific changes, and then to “train” AI databases to respond to the signals that indicate undesirable tissue changes.

“I don't think Dual Laser by itself will identify all the changes we are interested in, but the current methodologies in the industry don't do it either. However, I think when you blend them together with the right level of data model training, we can start to create some magic”, he says.

“Imagine an apple rotating as it moves past a camera. We hit it with custom multi-spectral light and the sets of images going to a data repository. Then that apple continues through the sensing rig to be hit with the Dual Laser, which allows us to develop another set data. Then we push those two data sets together through an AI engine to do the classifications we’re training it to do.”

Destructive testing of scanned items is done to make sure the training is based on the true values they are wanting to identify and so that any classification is accurate. The team is currently working on onions, apples and kiwifruit.

Says Geoff: “We haven't done all the testing work yet. We've developed the multi-spectral work over the last 12 months, and we’ve only had access to the Dual Laser for a few months, so that means we're working out how to fit it into a sensor that can be used in a high-speed grading line. That’s a reasonable chunk of work by itself and then working ahead and merging it with the other technologies is another sizable chunk of work.”

Ultimately the company expects to come up with a range of sensing technologies that each individual piece of fruit will pass through, and the results will be integrated and a decision made whether it's acceptable or not for a specific market or give it some sort of grading value that will indicate where it could be best marketed.

This is somewhat different to what's been done in the past, in that the system identifies numerous characteristics that can allow the grader, the farmer or the packhouse to optimise the use of the crop. For example, there may be some small defects that are not visible and don’t affect the shelf life, or nobody cares about, so the product can go into a grade for a specific low-end market.  Or perhaps this is a high-end product and needs to remain high quality for 45 days or more overseas shipment, so a more exacting set of grading criteria are applied.

“It's trying to identify a whole bunch of characteristics and attributes that that the end consumer actually puts value on and then helping optimize where that goes – different attributes for different markets,” says Geoff.

The company anticipates that the result will be a “bolt-on” technology that can be added to existing rapid grading systems, rather than needing a whole new processing line and thus keeping the cost down.