IoT, AI, ML: Keeping Innovation in Sight
Supply chain leaders eye up the Internet of Things, artificial intelligence, and machine learning to better predict demand, increase efficiency, cut costs, and power growth.
Robert Bosch LLC, more commonly known as Bosch, operates more than 270 manufacturing plants and thousands of manufacturing lines to produce millions of products. These include mobility solutions, consumer goods, and energy and building technology. Bosch management has set a goal of ensuring that, by 2025, all its products will either contain artificial intelligence (AI) or have been developed with the aid of AI.
Defining Moments
AI Initiatives look to growth
Among other applications, Bosch is currently using AI to improve quality, plant logistics and efficiency, and to reduce costs. “Our use of AI is widespread,” says Rahul Kapoor, vice president and global head of AI, Bosch Center for Artificial Intelligence.
For instance, Bosch is using AI to improve intraplant logistics, or “milk runs,” in which materials move from warehouses to production lines to support just-in-time manufacturing. Previously, the materials were shuttled at fixed times on fixed routes.
While that worked well, the schedules created with AI help manage demand changes and introduce flexibility. “The solution is helpful to optimize and dynamically adjust routes, so we only deliver and pick up material where it’s truly needed,” says Andy Hassold, connected industry consultant with Bosch Rexroth, a subsidiary of Bosch.
Bosch also combines a manually developed sales and demand forecast with one produced using machine learning (ML). “We’ve seen improvement by combining the methods,” Kapoor says.
Supply chain professionals across numerous verticals are implementing technology solutions that incorporate AI and its subset ML, as well as the Internet of Things (IoT). While the pandemic may have slowed investments, it hasn’t stopped them. About two-thirds of organizations responding to a recent Gartner survey either didn’t change or increased their planned AI investments since the onset of the pandemic (see sidebar, page 170).
In another survey, IDC found that growth in spending on IoT slowed in 2020, but was still up 8.2%. IDC predicts a return to double-digit growth in 2021.
Driving Growth
What’s behind the sustained investments? “It’s a combination of AI, IoT, and ML, now turbocharged by 5G, that is revolutionizing supply chains,” says Warren Chaisatien, global director of IoT customer success marketing with Ericsson, a Swedish networking and telecommunications company. Among other capabilities, the technologies can analyze data, including real-time sales and customer preferences, to better predict demand. In warehousing, AI-powered visual inspection can identify and correct damage before products are shipped out.
Among the verticals implementing these technologies are retail, manufacturing, and food production. One common thread across these supply chains: “They’re no longer linear,” says Drew Ehlers, global futurist and general manager of SmartPack, aload planning solution from Zebra Technologies.
These networks replace supply chains, and distribution centers, warehouses, manufacturing plants, and retail stores all become fulfillment nodes. “Fulfillment happens dynamically across the network,” Ehlers adds.
Ensuring Food Safety
In food production and distribution, these technologies can ensure safety and quality. Schwan’s Home Delivery, a leading direct-to-home food delivery company, operates 300-plus depots and more than 3,000 trucks, all of which need to be kept at specific temperatures—well into negative numbers—to ensure food safety and quality. Until recently, employees manually checked and recorded the temperatures, including after hours and on weekends.
No more. Schwan’s is implementing wireless sensors within the trucks and freezers that continually monitor temperatures and transmit the findings to both drivers and management. “I can see the freezer temperatures on all trucks at a glance,” says Larry Gaskin, national director of warehouse operations. “It provides a level of confidence.”
As of mid-October 2020, about one-third of Schwan’s trucks and depots had been outfitted with the SmartSense solution from Digi International, a provider of IoT technology.
Installing the sensors is easy enough that each depot can do it themselves. And when the company retires a truck, it can simply take out the existing equipment and move it to another truck. “It’s easy to install and follow up on,” Gaskin adds.
AI can also help food producers and retailers quickly establish the provenance of an item, says Rob Bailey, chief executive officer and founder of BackboneAI, a data automation company. Using AI, data on the farm where a piece of produce was grown can be mapped to the destination data at the retailer, and then probabilistically identify the source of items that might be contaminated—all within minutes. Historically, this process has taken weeks.
These technologies may even improve the global food supply chain. In 2019, IBM and Yara, a crop nutrition company, partnered with the goal of increasing food production on existing farmland by applying AI, ML, and big data within a global digital farming platform. This platform will integrate with IBM Food Trust, a blockchain-enabled network of food chain players. Ultimately, the new platform will enable “greater traceability and supply chain efficiency,” the companies say.
Retail and CPG BENEFITS
While the pandemic has hammered many brick-and-mortar stores, it also accelerated the shift to online shopping. That’s prompting many retailers to invest in AI for several reasons, including to improve inventory and supply chain management and better predict demand, according to Meticulous Research.
For instance, AI can assemble and analyze data from point-of-sale and inventory systems and identify patterns. Then, it would let a retailer know that instead of the dozen cases of Coke it typically sells in one shift, it only sold six, for example.
With this information, users can determine what’s behind the change and take action, including adjusting their supply chains.
Similarly, consumer product companies can deploy AI to read demand and then adjust their supply chains to optimize what they sell to each retailer, says Dinand Tinholt, vice president of insights and data for consulting firm Capgemini North America. AI can also help them repurpose machines to reflect changes in demand.
Retailers and CPG firms can also use AI to personalize products. Nike Fit, for instance, uses a mix of technologies, including data science, ML, and AI, to collect more than one dozen data points from customers’ feet. It can then recommend the shoe size likely to best fit them.
Not only does this improve customer service, but it can also reduce returns. The top reason for online returns in 2019 was incorrect size, fit, or color, found “The State of Online Returns,” a Narvar report.
Manufacturing Revolution?
Like Bosch, many manufacturers are using, or are interested in using, AI and IoT to improve production and distribution. “We’ve seen an acceleration of digitization,” says Rohit Gupta, chief executive officer and co-founder of Auditoria.AI.
Some manufacturers are connecting machines and equipment with IoT and then applying AI to more precisely schedule maintenance. “In the old days, you’d just run a machine until it broke down,” says Alan Salton, director of innovation with Panorama Consulting Group.
Or, a plant manager might set rather arbitrary maintenance schedules—say, performing maintenance after 300 runs—whether or not it was needed. By alerting management to upcoming maintenance needs based on a machine’s operation, AI can reduce machine downtime and minimize scheduling disruptions.
As Bosch is doing, manufacturers can deploy AI for demand and supply planning. With AI, they can more accurately determine what and when to order from vendors, and whether they should, for instance, move products from Illinois to Arizona to capitalize on a demand uptick there.
The potential from combining IoT, 5G, AI, and ML within manufacturing will likely lead to “a renaissance of innovation in the entire manufacturing space in the coming year,” Tinholt says.
While AI, IoT, and ML hold great promise, a few steps can boost an implementation’s effectiveness.
-
- Start with a strong business case. “Don’t just go after the shiny stuff, but understand what the technology can do for the company and its processes,” Hassold says.
- Clean and update data before starting. Without accurate, updated data, any predictions are apt to be less reliable.
- Keep in mind that the greatest value comes from solutions that provide insight to the future. “Lots of companies can report what happened,” Tinholt says. “You need to bring in what’s going to happen.”
- Look for solutions that minimize reliance on the IT staff—which often is already stretched thin—or that require a background in data science, advises Gary Brotman, vice president of product management and marketing with Secondmind, a provider of AI-powered decision-making platforms. At the same time, companies “should do their homework and learn what’s under the hood” of any decision-making solution, Brotman suggests. That is, they should understand why the system recommended one action over another.
- Incorporate security from the initial design, rather than view it as the figurative “cherry on top of the cake,” Ehlers says.
Risks OF Waiting
The challenges of implementing these technologies are real, yet inaction also carries risks. Companies that move forward with implementations are seeing competitive advantages, whether from operational efficiencies or developing the ability to add new features into their products.
Because the technologies continue to advance, firms that wait risk being left behind. Says Hassold: “The longer you wait, the farther you’ll drift apart.”
Defining Moments
Just what do the acronyms AI, ML and IoT mean? While it’s difficult to find a universally accepted, single definition, the following are reasonable starting points.
Artificial intelligence (AI): Refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions, according to Investopedia.
Machine learning (ML): A subset of AI. Refers to teaching machines to make predictions without explicitly programming them. Instead, they use historical data to predict new outcomes.
Internet of Things (IoT): IoT refers to connecting devices and capturing input from multiple sources.
AI Initiatives look to growth
Twenty-four percent of respondents’ organizations increased their artificial intelligence (AI) investments and 42% kept them unchanged since the onset ofCOVID-19, according to the results of a September 2020 Gartner poll of roughly 200 business and IT professionals.
Growth—namely customer experience and retention as well as revenue—along with cost optimization were the top focus areas for their current AI initiatives.
Over the course of the next six to nine months, 75% of respondents will continue or start new AI initiatives as they move into the “renew” phase of their organization’s post-pandemicreset.
“Enterprise investment in AI has continued unabated despite the crisis,”saysFrances Karamouzis, group chief of research at Gartner. “However, the most significant struggle of moving AI initiatives into production is the inability for organizations to connect those investments back to business value.”
While 79% of respondents say their organizations are exploring or piloting AI projects, only 21% state their AI initiatives are in production.
Lack of AI Talent is a Myth
The minimal strides made across organizations in operationalizing AI cannot necessarily be attributed to a lack of AI talent. Only 7% of respondents said that limited AI skills are a barrier to AI implementation, finds a Gartner survey of 607 IT leaders in November and December 2019. Instead, security and privacy concerns, along with the complexity of integrating AI within existing infrastructure, are at the top of the list.
“AI talent is not one thing; it’s multiple things,” says Erick Brethenoux, research vice president at Gartner. “The biggest misconception in the journey to successfully scaling AI is the search for ‘unicorns,’ or the perfect combination of AI, business, and IT skills all present in a single resource.
“Since this is impossible to fulfill, focus instead on bringing together a balanced combination of such skills to ensure results,” he says.
Organizations with the lowest AI maturity level are not experiencing a shortage of AI skills, with 56% reporting they either have enough talent or can easily hire or train AI talent. As organizations increase in AI maturity, so too does the level of reported AI talent, with 89% having no issues acquiring AI skills at the highest maturity level.