Predictive analytics is already transforming the transportation world. Artificial intelligence (AI) and machine learning (ML) have the power to make fleets more productive and profitable. But it still takes people to get the most of these emerging technologies.
That was the overreaching message of a recent Trimble in.sight virtual panel discussion on new supply chain trends and how some businesses embrace AI technology to improve efficiencies and make better decisions.
A few years ago, the buzz in the transportation industry was that “AI was going to solve everything,” according to David Dunst, IT manager at Paper Transport. He expects AI to take off over the next five years—but it won’t take over the trucking industry.
“AI really adds value when it’s augmented intelligence,” Dunst explained. “It’s about providing guidance to the human, who then makes better, faster, more informed decisions. When we take all the knowledge we have as humans and apply AI—when we put those together—we become so much more intelligent and so much better at what we do as businesses.”
AI and ML “fall under the umbrella of predictive analytics, which predict what the future will look like,” said Peter Covach, industry solutions adviser for Trimble, who moderated the June 24 panel. “More interestingly, AI and ML have a more immediate use in predicting what the computer is seeing.”
Covach said this could be as simple as a social media site automatically identifying and tagging a person in a photograph or as complex as a self-driving vehicle predicting when it should apply its brakes. “The computer is interpreting the situation based on the data you give it,” Covach explained. “By using advanced mathematics and historical data, data scientists are able to make those predictions with a high degree of certainty. These same concepts that power self-driving cars or recognize faces can be used to fuel business objectives to the next level.”
Better business decisions
Transportation and logistics company Covenant Transport Services uses Domo, a business intelligence and analytics tool that transforms data into live visualizations and real-time metrics. Matt Mullins, VP of program management for Covenant, said the AI solution was easy to implement in his fleet.
Domo provides Covenant workers feedback to help make more informed decisions throughout the day, Mullins said. “This could apply to transportation for resource utilization for a specific market; it could be the need to move resources around better for better utilization of picking and warehouses—we’re 3PL (third-party logistics) as well,” he said. “So maybe we need to determine if we want to change a slotting of a specific SKU (stock-keeping unit), based on trends, for picking.”
Mullins explained that the Domo program helps Covenant “see across the warehouses” to help the fleet slot freight and equipment across its supply chain. He said the company plans to expand the AI usage that comes with the business tool. “Right now, we use it as an analytical, slightly predictive tool. We’ll get to the point where we’re much more predictive; then we’ll add some machine learning along the way at some point.”
For now, Covenant’s people are still making the big decisions. “But there are several areas that we’re allowing it to make decisions based on those more mundane tasks,” Mullins said.
Robotic process automation
Some of those mundane tasks are completed using robotic process automation (RPA) for repetitive, labor-intensive needs so humans can “utilize soft skills, which humans are meant to be doing,” Mullins added.
For-hire truckload carrier Paper Transport has also embraced RPA. “We found that we had a bunch of processes that users were just consistently doing the same thing day in and day out,” explained Dunst, the fleet’s IT manager.
“We know we have the information on our systems on what they have to do,” Dunst said of planners at Paper Transport. “So rather than having them going to these websites and setting appointments, or searching for available loads to backfill our network, we use RPA tools to actually go into our customers’ sites and set the appointment that we want. We can search broker sites and find what loads are available and then bring all that information back to the user or push that data to the customer or to our systems to further the delivery of the load.”
Dunst added that RPA helped Paper Transport grow as a carrier without hiring more planners and office personal. “The process we had before—where people were having to do all this manual work—meant we would just have to keep hiring and hiring and hiring to keep up. That’s not sustainable if you want to grow as an organization. We found that RPA really took away those tasks that people didn’t want to have to do and open them up to take on some more meaningful and valuable work.”
AI and humans
Paper Transport first “dipped its toe into AI” to better understand which drivers in its fleet are more accident-prone on the road, Dunst said. “It allowed us to build a model that helps us understand who are the drivers that might be at risk and how do we change our behaviors with those drivers to ensure that those drivers don’t have an accident or an incident,” he noted. “That was the first thing we ever did with AI.”
Dunst said that its first foray into AI helped the fleet reduce significant vehicle accidents because it led fleet managers to engage more with drivers. Along with making the Paper Transport fleet safer, it helped with driver retention “because drivers felt more personally engaged with the company,” he said. “We were directly reaching out to some of those drivers and saying, ‘Hey, how’s it going? What’s going on with you?’”
That work with AI led Paper Transport to Trimble’s Dispatch Advisor, an optimized dispatch solution for the transportation industry that prescriptively suggests matches between loads, drivers, and equipment.
While the accident prevention work was “much more on our shoulders” as a fleet, Dunst noted, the dispatch solution was “more of a partnership, where Trimble brought a bunch of data scientists, resources, and knowledge to the project.”
Trimble Dispatch Advisor reviews all available matches by looking at variables, including location, drivers’ hours of service data, time windows, and more, such as the effects of selecting a specific load on future orders. The system then recommends a primary match, along with reasons for the recommendation. Other matches are also suggested, including an estimated impact on the company should they choose any other options rather than the first or primary match.
This, Dunst said, made “a huge difference from a profitability standpoint because the Dispatch Advisor was bringing up loads in front of our planner that they didn’t even realize would be a good fit for that driver.”
He said his planners embraced the guidance of the AI product. The Trimble solution led the fleet’s planners to take less time to pick a load. The loads they were selecting reduced the miles drivers needed to cover to pick up or deliver freight. “That really positioned our drivers to get home on time, and it cut down on some of the additional costs that we were seeing previously,” Dunst said.
The more fleet data available, the more powerful and accurate AI-driver solutions become, said Chris Orban, VP of data science at Trimble. He also noted that AI solutions are there to assist humans in decision making, not necessarily make all the decisions.
“As powerful and as cool as these terms AI and machine learning are, we still as humans might know something that the computer doesn’t,” Orban said. “The computer can process huge amounts of information very, very quickly. But our human brains might know that a particular driver really hates to run in Pennsylvania, or they’ve never driven in snow before. So before I send them over Donner Pass in the winter, let’s figure some other plan out. Those are the kinds of things that we need to be able to supplement with our human knowledge even as we get more advanced on the machine learning side of things.”
