From Manual Labor to High-Throughput Automated Internal Quality Analysis
How long do you inspect an apple at the store before it ends up in your basket? How much time is spent determining the apple’s quality before it makes it to the store? Today’s consumers expect their food products to be high quality -- and consistently so. Fortunately, companies such as Ocean Optics are using advanced optical sensing tools and analytics to improve the quality of food for consumers and the quality of business for food processors and sorting machine manufacturers.
Traditionally, food sorting has been managed manually, relying on the expertise of workers to make some judgment of quality. The introduction of optical scanning and spectroscopy has added a much deeper level of understanding and analysis, with some instruments being able to see inside a fruit peel or determine exact fat or water content.
Ocean Optics has provided world-class spectroscopy tools to industrial customers for decades, and has much experience in traditional food analysis approaches. However, there are some new tools in the Ocean repertoire that are a complete game-changer: Machine Learning and Artificial Intelligence. By fusing traditional spectroscopy with advanced statistical models and machine learning architecture, immediate benefits can be realized: for integrators seeking more advanced offerings, for processors wanting more efficient facilities, and for consumers wanting to be confident they are eating what they expect.
Date Fruit Sorting Case Study: Getting Started
Dates have been a popular fruit since biblical times. Consumers know exactly the kind of date they want to buy: not too wet and not to dry.
Ocean Optics was approached by Lugo Machinery & Innovation, a leading supplier of fresh produce sorting products based in Israel, to improve their manual method of sorting dates by moisture. Their goals were simple: Automate the sorting process to eliminate all manual inspection, and perform the measurements rapidly and non-destructively. Additionally, Lugo’s timeline was very short, with only four months until date season, and they had no prior experience with spectroscopy.
The Lugo-Ocean Optics Partnership Bears Fruit
Ofir Luk, one of the co-founders of Lugo, shares his insight on the challenges of fruit quality sorting and the company’s experience working with Ocean Optics:
"Lugo was one of the first companies to develop sorting technology for dates, in 2003. Our technology is unique in structure and work method.
The global date market continues to grow and poses technical challenges for quality sorting. We develop new technologies for identifying internal quality and defects of fruit. The search for innovative technology led to cooperation with Ocean Optics.
The fruitful collaboration [between Lugo and Ocean Optics] began with us sending samples for the first examination and has included regular meetings with the technical teams from both companies. Sharing knowledge in optics, mechanics and software brought us, in a short time, to the desired result.
We are about to launch a first system, so that we can offer technological solutions to our customers that will meet their needs and provide full value for their investment.
We found in Ocean Optics a determined and talented team that provides us with real-time answers."
Lugo provided date samples for feasibility testing, which quickly showed NIR correlations to moisture levels in the fruit; this analysis was used to determine the best optical hardware components for the system. This optimized system was taken to the customer site to perform analysis on a much larger sample set, which was used as training data to develop proprietary machine learning algorithms immune to interferences experienced by traditional chemometrics.
Lugo was aware of date moisture response between 850-900 nm and imagined the analysis would focus solely in that region. However, Ocean extended this spectral range to include other repeatable spectral features that held value in the statistical processing. This level of applied spectral knowledge is something Ocean Optics can offer to customers who otherwise would not have access to such insight.
Date Fruit Sorting Case Study: Advancing Understanding
From the first computer learning program written by Arthur Samuel in 1952, which simply learned which checkers moves were part of winning strategies, the deeper potential of computing has been recognized. Computers are not just glorified calculators, always yielding the same result from the same inputs, but instead are malleable and “living” tools that can make clear observations about the world and use those observations to make logical decisions relevant to current conditions.
At the industry level today, we see technologies like Watson from IBM changing the way self-learning models are deployed into production at scale … whenever they’re not busy beating human Jeopardy! champions. A year after Watson’s historic Jeopardy! win the market saw more consumer-level technologies come about, including several Google algorithms from their “X Lab” that could browse the Internet and identify specific objects autonomously. Facebook and Amazon soon followed suit with DeepFace and AWS-Machine Learning platforms, respectively, and personal assistants with these underpinnings are finding their way into every aspect of life -- from your phone to your car and even to home appliances such as your oven and washing machine.
Applied analytics also factored into the Lugo date sorting project. Having shown the robustness of the algorithms developed earlier, as well as their ability to predict with such high accuracy, Lugo integrated their spectral platform into a conveyor-belt system with the algorithms running on a devoted PC. The architecture of this sorting system scans each date and weighs over 12 potential correlation models, ultimately “voting” on the best model and generating the output in milliseconds. Classification methods were valuable in this application, which required broader threshold decisions versus precise numerical outputs; some of these methods include k-nearest neighbors (k-NN), Gaussian, and polynomial. The method toolbox is growing to include regression models for accurate quantified outputs, which is invaluable to those working in process streams looking for impurities from any number of sources.
Regression analysis has been limited by the human factor since its inception, relying on what was visible to observers and their knowledge of the system in question. Classic chemometrics in spectroscopy starts with everyone’s most memorable high school chemistry lesson: Beer’s Law.
By drawing mathematical correlations between absorbance and concentration trends we have taken the first step toward advanced understanding. As these correlations evolve from linear fits to more complex functions, this moves further along, and as these complex functions begin to account for multiple species, we move further still. But at some point, a wall is hit; at some point there are so many interconnected inputs working to generate so many outputs that traditional deconvolution methods become daunting if not impossible.
Introducing machine learning to the fruit sorting process brings a deeper level of analysis that is not relying on the human eye and perception but is rather folding its fingers through every digital bit of data, statistically analyzing the entire data array of each spectrometer scan at every pixel on the detector. This system can now see abnormalities previously missed and develop smarter, more streamlined correlations for far more variables intertwined with one another.
One of the most powerful aspects of machine learning is the ability of non-experts to introduce new training sets to the system and achieve results previously attainable only to experts in the field. This creates the freedom to develop beyond the initial product capabilities without the cost or stress of finding and paying an expert for such expansions and evolutions. With these new tools, what was once highly intimidating and restricted is now entirely approachable to many levels of users.
Taking the observation and making the subsequent decision are the most critical elements for Ocean Optics but are not the goals of the user. For the customer, an answer is the end goal. Indeed, translating the decision into some physical action is where the true value of the system is seen.
Ocean Optics offers a range of communication interfaces to work with PLCs and common process equipment. In our case study, Lugo and Ocean Optics engineers worked to have the spectrometer module communicate with the system PLC for triggering action to a valve that steered the dates in the right direction (appropriate moisture category). This functioning system created a beta platform to refine and optimize the algorithms before final implementation.
Date Fruit Sorting Case Study: Applying Insight
Today, Lugo is using a fully integrated system that scans 5 dates per second, or every 200 milliseconds, and is entirely free of human intervention. This has reduced overhead cost and improved safety at the facilities that use the system, and sets a foundation for the future using advanced tools that can be further evolved.
Industries worldwide have trusted Ocean Optics spectroscopy systems for decades, but have been limited by their own spectral knowledge in extracting the maximum value from those systems. As technological trailblazing has been in our blood for decades, we have not hesitated in adopting the immense value of advanced statistical models and the ability of modern computers to make intelligent decisions about complex processes in fractions of a second.
Are you ready to bring your process into the new millennium using the tools and standards of tomorrow? Talk to us about the power of spectroscopy and what a partnership with Ocean Optics can do for you. After all, that apple isn’t going to judge itself.