Artificial intelligence (AI)-based technologies are just about everywhere these days, making electronic devices, equipment, and business processes more streamlined and, of course, smarter. As it is commonly understood, AI uses computers and machines to mimic the problem-solving and decision-making capabilities of the human mind—and it exists on many levels. Common examples include speech recognition, online virtual agents, computer vision, and even “recommendation engines”—those systems that tell you what “you may also like” when you’re shopping online.
AI is also being applied in the warehouse, in the form of emerging technologies designed to increase output, reduce errors, maximize equipment uptime, and help companies bridge the labor gap by accomplishing more work with fewer people. Here’s a look at some emerging applications in use at warehouses nationwide.
One of the newest terms to hit the warehouse floor is a warehouse management system “accelerator,” which is a software solution that sits above a company’s warehouse management system (WMS) to help optimize and orchestrate the broader operations of a warehouse, according to Keith Moore, CEO of AutoScheduler, a WMS accelerator founded in 2020.
“We are a fairly new breed of software,” Moore says. “We are a complementary solution to an existing WMS, and our objective is to be the overall brain for a warehouse. That’s the easiest way to put it.”
He explains that the strength of a WMS lies in its ability to integrate with the hardware and robotics systems on the warehouse floor to manage inventory, coordinate the picking and packing processes, generate analytics, and the like. But the same system may come up short when it comes to optimizing the various constraints at work in the warehouse—labor, for one—so that managers can efficiently execute all of the disparate functions and maximize overall flow through the facility, he says.
“For example, [a WMS] would struggle to leverage data to understand when a trailer arrives, the optimal door to put it in when it gets there, what should be cross-docked, and so forth,” Moore explains. “If you have 80 people working, what equipment do they need to be on, what tasks should they be working on, how do you maximize flow through the building? We try to provide that total plan for execution.”
AutoScheduler accomplishes that goal by combining artificial intelligence with digital-twin technology to model the workflows throughout the entire warehouse or facility complex. The digital twin models those flows, looking ahead 48 to 72 hours; AI is applied on top of the twin to calculate and determine the best sequence for all of the processes that need to take place in that time period. It sounds simple, Moore says, but it’s anything but: Warehouses are dynamic workplaces, which means the accelerator is constantly calculating and recalculating to optimize process flows. AI is the brains behind it all, continuously working through math problems with far more dimensions than a human could ever handle.
Moore likens the situation to a game of chess in which the WMS accelerator is the ultimate player.
“It’s not possible for a person to consider all the possible combinations in chess—and warehousing is more complex than chess,” he says. “[The AI is] constantly re-planning and finding the best way to optimize. It’s a constant reoptimization of the total system. That’s the differentiator. And it allows you to see ahead and plan better.”
Moore says larger companies—those with multi-building campuses—benefit the most from WMS accelerator technology but adds that smaller facilities with 15 to 20 people can benefit as well. Some of the most common improvements include higher fill rates, increased output per hour, and a reduction in detention and demurrage costs.
“The number one thing we are doing is enabling delivery of products to customers. The warehouse should never be a bottleneck; it needs to enable the flow of products through the supply chain,” Moore says, adding that demand for WMS accelerator technology is poised for growth over the next few years as organizations place a greater emphasis on the warehouse in general. “Warehousing is finally getting a bit of a moment in the spotlight. It’s always been a cost center, but companies are starting to realize that having a really effective warehousing innovation strategy can become a significant differentiator.”
AutoScheduler counts Procter and Gamble and a host of other large consumer packaged goods (CPG) companies among its growing list of customers.
AI-powered image recognition tools are another example of cutting-edge technologies that are improving operations in the warehouse. Software company Siena Analytics is applying the technology to high-volume logistics operations—to increase throughput and efficiency, and also for quality improvement, according to company founder and CEO John Dwinell. The company’s Siena Insights software captures data from the sensors found in package-scanning tunnels and sorting equipment in the warehouse; it then analyzes that information to identify equipment problems and assembly-line bottleneck, as well as labeling and packaging issues that may lead to delivery errors or quality-control problems. The company analyzes data from millions of packages flowing through warehouses daily.
“We’re using AI to ‘see’ every one of those packages,” Dwinell explains. “You can train the AI to look for all sorts of different features and report back on every single package … [which is] good for throughput and efficiency as well as for product [quality] and compliance. And those are really big topics for anyone’s logistics operation.”
As Dwinell explains, Siena Insights is vendor-agnostic, meaning that it can analyze data from any brand of scanner, sensor, or camera to provide a standard solution for improving package flow—which includes identifying problems such as incorrect packaging, a misapplied or missing label, product damage, and the like.
AI-based “deep learning” technology is at the heart of the solution. A subset of machine learning, deep learning teaches computers to learn by example, using large amounts of data and artificial neural networks that contain multiple layers. It’s the technology behind driverless cars, and it also powers the voice-control features in cellphones, tablets, and other consumer devices. Applied to scanning and image recognition in the warehouse, it provides real-time visibility into the package’s journey and its condition along the way.
“We’re using deep learning models to look at the images, and they are trained to identify all kinds of features: the type of packaging, its condition—are the labels there or not there?” Dwinell says. “Is the package wrapped in plastic? Does it have an open top? Does it have a crushed corner? How are the bar codes? Are they readable? If not, why? We train the models to recognize these features and let them run in real time.”
Warehouse managers can then use the data to make process improvements, monitor equipment health, and automate sorting and exception handling—all of which leads to higher productivity and better quality.
“What we’re really bringing [to customers] are solutions for compliance and quality in general,” Dwinell says. “We are identifying for them where things are right and where things are wrong, and then showing them what is wrong—and we’re not just giving them the information; we’re providing a picture to go with it.”
Siena Analytics works with large, high-volume logistics and supply chain companies, including third-party logistics service providers (3PLs).
Industrial battery and energy solutions provider EnerSys is using AI to help its customers find the best solution for their application—a factor that varies from warehouse to warehouse, according to Kerry Phillips, the company’s vice president, global product management, motive power. EnerSys uses its EnSite simulation software program to analyze battery and equipment data—which are gathered by interviewing the customer and extracting data from an electronic device EnerSys attaches to the customer’s material handling equipment. The software uses simple AI-based algorithms to calculate which energy solution is best for the job, taking into account anticipated changes that may require a different solution down the road.
“We’re trying to understand the customer’s application and select the right product,” Phillips explains, noting that that could range from a traditional lead-acid battery solution to a more advanced lithium-ion product. “The really cool thing is, if we think the customer’s business might change—for example, it may go from one shift to two shifts, or its break times may change—we can say, based on that growth, this is what it should use or this is when it should switch to a new solution.”
Phillips says the EnSite program differs from traditional battery-management systems in its predictive simulation capabilities. That is, it doesn’t just analyze the demands being placed on a particular product for maintenance or energy-savings purposes, but can also suggest a range of solutions that could better meet the customer’s particular needs. It even includes a financial model that takes into account a product’s purchase price, maintenance costs, and a host of other factors over the life of the battery so that the customer can weigh different options.
“We have some customers that have chosen a virtually maintenance-free product because of water consumption, for example,” Phillips explains. “In places where [water consumption] is a big issue … that is a cost savings, an environmental savings, and a labor savings. EnSite can factor all of that in.”
Phillips says EnerSys is working on a more advanced predictive analytics tool that will take the simulation software to the next level.
“The next generation is really where the AI will become progressive,” he says. “We will be able to predict end-of-life and then project out what [a customer will] need next. As we evolve our use of AI, we will also be able to predict service needs so we can optimize the service life [of a product].”
Phillips adds that no matter where AI is being applied in the warehouse, the goal is to get a glimpse of the future so that managers and workers on the floor can make better long-term decisions.
“I think the trends you see in the battery industry are the same as you see in other industrial products,” he says. “We’ve got to have smart products … and all of that can be achieved through the use of data and how we report that data back.”