Watch Now


Industry Insider: Seeing the future of logistics

Nascent Technology looks to take image analytics from the factory floor to the terminal gate.

   In 2012, Google announced to the world that it had successfully given a computer a set of eyes.
   What had the computer seen? Cats. With 1.8 billion images shared daily on the Internet, pictures of cats are overwhelmingly the most popular. In fact, the computer had seen so many cats that it had developed pattern recognition.
   Since that discovery, the use of image analytics has grown exponentially in every field in the past few years, including port and terminal technology.
   “Image analytics” may sound like yet another intimidating technological buzzword, but the term has been around since the 1940s, first used in the context of quality control in manufacturing.
   Back then, manufacturers had optical comparators, giant lenses that magnified each pristine new product, so employees could carefully inspect for imperfections. As the technology evolved, those machines gave way to cameras.
   Now, Nascent Technology LLC, a provider of automated gate systems for port terminals, aims to become the “vision experts” in logistics technology.
   “When it comes to the future of port and terminal velocity, the goals aren’t that far removed from the software that enables Pepsi to remove a damaged bottle from their production line,” explains Chris Kete, CEO at Nascent. “We are developing software now that is able to distinguish a UPS truck from a FedEx truck or a J.B. Hunt rig. The software accomplishes that by holistically capturing a myriad of conveyance information that goes well beyond reading box numbers and license plates. The technology takes into account everything from shape to color to numbers, even initials on the tractor.”
   But identification is only one objective. Imaging and optical character recognition (OCR) portals already are fairly common throughout the industry and have allowed most port terminal customers to remove inspectors from the lanes. The inspectors, however, are still required to check for damages remotely.
   Although challenges remain in terms of the required hardware and software, Nascent currently is researching, developing and testing image analytics technology for its enVision portal that has the potential to automate those identification and inspection processes entirely.
   “Speed is the revenue of logistics, and yet currently a lot of time is lost entering and exiting a facility because there are a lot boxes that have to be checked — like hazard placard inspection or damage inspection,” said Kete. “Imagine instead a driver passing through a computer portal that will do that automatically in real time.
   “You accomplish so many goals,” he said. “You’re checking for damages, so there are fewer liability issues, and exponentially increasing velocity, thereby increasing revenue for your facility and your customers. They can get their freight in and out faster.”

Research & Development. Nascent’s secret weapon in developing vision software is Chris Dahms, the computer vision engineer who moonlights as a popular vision software trainer on YouTube. His first feat for Nascent has been putting hazard placard identification into production at two companies.
   The placards identify classes and subclasses of hazardous materials, denoted by both color and design. Utilizing pattern recognition and OCR, Nascent’s hazard placard identification system can identify whether all placards on the conveyance match for the sake of compliance, as well as direct stowage, given the requirements of the hazard class. This feature is of particular use to ports, as certain materials cannot be stowed below deck, while others cannot be placed in close proximity to one another.
   How accurate is the software?
   “On a routine day, we’re approaching 100 percent accuracy,” says Dahms. “We generally read ISO containers around 96 percent. Weather is a consideration. A snowy day might result in a few more lower confidence reads — around 89 percent — as some placards will likely be dirty. That amounts to two or three containers out of 100 that require human intervention, as opposed to all 100. But another feature of the system is the threshold for accuracy is determined by the customer, rather than us. It’s adjustable.”
   Other technologies under Dahms’ purview include:
     • Carrier identification, the real-time verification of high-volume logistics company trucks such as UPS and FedEx to move more quickly in and out of the terminal;
     • Automatic damage detection: Currently conducted by remote operators, enVision’s image analytics will soon be able to detect holes, dents, bent landing gear and broken rods;
     • And tire wear detection: A frequent sticking point between terminal and freight companies, Nascent’s vision experts currently are testing an image analytics application that can detect not only damages but anomalies such as underinflated tires, poor tread depth and sidewall damage.
   As the enVision product evolves and its vision capabilities increase, Nascent is exploring practical uses for the technology in customer terminals.
   For example, the common need for detecting the presence of a vehicle, which traditionally has been met through the use of ground induction loops, can be addressed in a more cost-effective and efficient manner through the application of vision technology.
   “The problem with ground loops is the weather ages them, they have to be recalibrated, they break and they’re expensive to install, especially in existing installations because they have to be cut in,” explained Kete. “That means the lane is out of commission during surgery, and, again, speed is revenue.
   “So we are applying image analytics to eliminate the need for ground loops,” he said. “That’s an immediate benefit to our customers. Besides, our vision software can do things that loops can’t. It can ‘recognize’ the type of equipment that may be in tow.”
   There’s potency in this prospect, according to Kete, because the system can readily be designed to encompass the entire yard. The system can distinguish a truck carrying a 40-foot container from one with two, 20-foot containers on a 40-foot chassis or a truck with a bare chassis. The system then can classify things like what direction the driver is going, how fast and when he left.
   The implication is that a port or terminal could use this information to study yard traffic to a degree of accuracy that hasn’t previously been attained. The “big data” and analytics applications are more abstract but full of opportunities to increase yard efficiency.
   There’s a hardware component to the system as well that entails its own set of unique challenges.

Hardware Challenges. Chris Austin, Nascent’s automation systems engineer, noted that unlike previous applications of image analytics on a factory floor, the enVision system will operate outdoors.
   “That means there will be changes in light, which will affect the computer’s vision just as it affects ours,” he explained. “We don’t even realize what our brains are processing in a glare. When you take a photo of a shiny object in the sunshine, you get spectral highlights, which means there are no colors, just white space. Our mind fills in the blank automatically because you know what that object is. A machine won’t naturally do that, you have to teach it to do that, and that means you need the cameras and filters that will work in concert with one another to minimize variations and allow it to learn.”
   That means adjustments for the camera’s lenses, filters and software — in a word, “everything.”
   “I have to break this down into levels,” said Austin. “First, how do I get the light right for the OCR engine? Second, I have a light bank out there that has to withstand weather extremes. I’ve gone through eight cameras in testing and hundreds of permutations in filters and light.
   “The first phase of this testing is always in our lab so that the software can work from it. But eventually I have to take it to a production environment where the filter and light combination can be tested for anywhere from days to weeks to months, adjusting and tweaking.”
   And it’s not just sunlight that will affect the system. Austin noted inclement weather can be a major factor as well.
   “Extreme temperatures, rain, snow, fog and dirt are not optimal for your laptop or your personal camera,” he said. “The same is true for the cameras you put on a yard. So those must be durable and simultaneously work in perfect concert with the light, filters and software.”
   With two yards already in production on hazard placard identification, it’s safe to say prospects for Nascent’s enVision portal are “looking up.”
   “It’s been a long road,” said Austin. “But we’re finally seeing all of our hard work and testing pay dividends — not only for us but, for our customers. They are now able to gain huge benefits from the technology. You might even say, ‘Seeing is believing.'”

   Jeff Necciai is senior vice president of sales and marketing at Nascent Technology. He can be reached by email at [email protected].