By Jeff Erickson | April 2021

Judging by the worldwide reaction to a lone ship wedged across the Suez canal, you can see that the flow of global commerce leaves little room for error.

While most mishaps aren’t this kind of headline news, global ports, shipping terminals, and border crossings are one slip away from becoming the new bottleneck. They’re turning to software makers like Finland’s Visy, a global leader in optical character recognition (OCR) systems, to keep goods flowing using cameras, sensors, and ever more intelligent software.

Here’s how: Picture a gargantuan ship, stacked with containers 10 stories high, arriving in port to be unloaded, verified, sorted, stacked, and eventually moved onto waiting trucks or rail cars. It’s a complex dance, the faster the better, where timing—and time savings—are key. Visy’s camera-and-software systems recognize vehicles, open gates, and can even determine how a container is placed on a trailer as it enters the terminal, so the cranes and other container handling equipment start getting into perfect position before the load arrives.

“We can shave time off of each transaction by automating it using AI-based vision technology,” says Visy CEO Petri Granroth, whose software works by accurately reading and reacting to visual cues on the fly. Another example: Visy’s system can verify with near-perfect accuracy whether a shipping container’s 2-inch customs lock is present and even provide the lock classification.

Visy supports its software at large facilities around the world, and now also runs on Oracle Cloud Infrastructure (OCI), a move to help the company more quickly add features and scale up with large customers—and make it affordable to much smaller port and shipping terminal operators.

Reliability required

Born out of the University of Tampere, about two hours north of Helsinki, Visy began innovating with OCR more than two decades ago. Starting out as an early supplier of automatic gate systems for the Port of Helsinki, Visy has grown into one of the largest providers of gate automation for the shipping industry, with customers in 25 countries from Egypt to Indonesia to Brazil. That same expertise has helped it bring ingenious uses of OCR to border traffic control. From its early years of providing solutions to the Finnish Customs Agency in ports and at the Russian border, Visy has grown into one of the largest providers of border traffic control systems in the European Union.

The software automates gates and access-control systems, scheduling, and vehicle booking, and orchestrates movement of cargo throughout expansive facilities. With these automations, “a transaction that used to take two hours—from the time a truck hits the gates, goes in, picks up the box it needs, and then clears and leaves—shrinks to more like 30 minutes,” says Visy Vice President John Lund.

That also means that ports can’t afford for these systems to fail. “If our technology stops, it means that cargo stops moving at the customer’s site, and there are massive implications,” says Lund, in terms of labor costs and contracts based on meeting performance metrics.

Reliability is one big reason that Visy runs its systems using Oracle Cloud Infrastructure and Oracle Autonomous Transaction Processing. “The autonomous database is a cornerstone of the daily operations for a growing list of customers,” says CEO Granroth. The cloud database stores all vehicle, train, and container traffic data and connects to control systems where images are timestamped, stored, and processed by AI algorithms to make decisions on site.

“The flow of data is the air we breathe,” adds CEO Granroth. “Oracle, in our opinion, is the most stable platform that we could have for our system.” The data collected by Visy’s systems requires massive storage and compute capacity, and its Access GOS (gate operating system) with OCR software process more than 5 million critical events per day.

Visy wanted to move these workloads from an on-premises environment to the cloud in order to improve scalability, reduce infrastructure costs, and make its system more economically viable for smaller customers. “The Visy Access GOS and other data collection tools are an obvious choice for customers moving millions of TEUs a year,” he says. (“TEU” is an industry term for a standard shipping container.) But by using OCI and Oracle Autonomous Database, “we want to push that threshold down to make it financially feasible for small terminals, let’s say those at 30,000 TEU per year, to use the technology.”

As a growing independent software vendor (ISV), the company also needed to spend less time on database administration and maintenance, so that its 25 AI engineers can focus more on product innovation and customer support. “We needed a partner with a cloud database that is simple to use, maintain, and integrate,” says Granroth.

An engineer calibrates a camera that will help automate a gate using Visy’s AI-based vision technology.

Visy evaluated options from Oracle, Amazon Web Services, Microsoft, and other vendors, before choosing Oracle Autonomous Transaction Processing. “The performance was clearly the best, and we did not want any bottlenecks to our solutions,” Granroth says. Read more about Visy’s migration.

ISVs in a range of industries are increasingly choosing Oracle Cloud Infrastructure for the scalability, stability, and price-performance that Visy is seeing. Granroth encourages fellow ISVs to find partners “who fill gaps in your own capabilities and provide you with a launchpad for your products. We made the decision to work with Oracle early on, and it gave us the foundation we needed to provide solutions for our diverse customer base on a global scale.”

Reading visual cues

From the early days of reading license plates and other codes, “the advancement of vision technology over the past seven or eight years or so has been incredible,” says Lund. Visy has developed a deep neural network that can collect all kinds of data about cargo that helps port and shipping terminal operators run more efficiently, by anticipating what kind of loads are coming.

“Companies now want to know how many axles are on the truck as it’s coming in. They want to know where on the trailer the box is positioned, which helps them to make their cranes more efficient,” Lund says.

Remember those custom locks on the back of containers? Visy can inspect them while a container is moving, even though the locks are only two inches long and about a half-inch wide and come in several form factors. “The vision technology is so advanced that it can tell you on the fly whether it’s there and what type it is,” he says. If a human had to verify it manually, “you would have to be like three feet from it,” he says. “Because of the high pixel count of the cameras and the brains of the algorithm, we can tell you with 99.9% certainty whether or not that box is locked.”

Next, Visy is developing automatic damage detection tools to determine if damage to a container happened while it was at a facility, or if the damage was there when the container arrived. “This is truly going to be a disruptive technology for the industry,” says Lund, who notes that global insurance claims for damaged goods run in the billions.

“Of course, the question is always, “Who's liable?”” he says. “Now, we're able to train the algorithm to look for common types of damage as a box enters a facility. A customer can say, ‘This was damaged when it entered our facility and here's the proof.’” And because Visy’s AI flags any damage and either routes the container to a repair center or notifies a person for further inspection, “you don't need an army of people at the gates or scrolling through pictures all day,” Lund says. “If your terminal does 3 million boxes per year, well, you can do the math on how much time and money this will save.”

Photography: Visy