���ϳԹ���

Penske says it uses AI to supplement, not drive a decision or take over a process. The goal is not to reduce the number of supervisors but instead let them do higher-­value work. (Penske Truck Leasing)

[Stay on top of transportation news: .]

Increasingly in fleet maintenance, artificial intelligence has been used to guide technicians on repairs and diagnose problems proactively. It also has helped supervisors to better engage with and train their people more efficiently.

Kyle Kendall, vice president of maintenance analytics and systems for Penske Truck Leasing, says AI should be used as a tool, not a crutch.

“Humans are in control,” he said. “We’re enabling more insights through technology to help the humans be more in control and better understand.”

Penske, which ranks No. 13 on the Transport Topics Top 100 list of the largest for-hire carriers in North America and No. 16 on the TT 100 logistics companies list, partnered with a third party to build the system.

The company is leveraging data to drive preventive maintenance programs and performed more than 95,000 proactive repairs last year. It has developed scenarios and predicted times to failure. When a fault code is signaled, a Penske agent informs the customer. Ideally, the maintenance can occur when the driver is finished with the route or at a truck stop, or the repair can be matched with a scheduled preventive maintenance visit.

Improving Efficiency

Penske is now testing algorithms in numerous locations that will eliminate some of its supervisors’ paper and administrative work. The algorithms can understand written language and transcribe oral inputs on repair orders. The technology, officials say, will unchain supervisors from their desks so that they can be on the floor managing technicians and serving customers. The algorithms can serve as “another set of eyes” that can help supervisors make decisions, advise them on certain actions and make sure customers are accurately billed.

The fleet said the technology is meant to supplement, not drive a decision or take over a process. The goal is not to reduce the number of supervisors but instead let them do higher-­value work.

Image

Mangione��

“In my world, having that supervisor on the shop floor next to my technicians, that’s high value. Sitting behind a PC doing administrative work — not high value,” said Gregg Mangione, Penske’s executive vice president of maintenance.

AI also is utilized to point technicians toward the exact training needed based on user prompts. For example, a technician needing help changing a NOx sensor can be routed to the pertinent section of a video rather than forcing him to wade through an entire library — or AI can summarize the video.

Kendall said good data is the key to success, as AI can evaluate and clean up data as it’s entered. For example, the company has models that can evaluate when someone records an erroneous mileage amount.

Company officials expected resistance when they asked technicians to wear a headset and talk to the system as they did preventive maintenance. However, technician feedback has been positive.

Image

Kendall��

Penske has guardrails with tight controls, governance teams and an AI leadership committee, so there is always a “human in the loop.”

“We’re trying to engineer solutions to problems,” Kendall said. “We’re not in there trying to force AI down somebody’s throat.”

Other motor carriers are incorporating AI into their operations.

Ryder System, No. 6 on the for-hire TT100 and No. 8 on the logistics TT100, is laying the groundwork to integrate AI tools into its products, including the RyderGyde fleet management system, said Jenn Dixon, vice president for maintenance technology. “Additionally, we are collaborating with our OEMs to assess their AI advancements, ensuring we maximize their innovations to stay at the forefront of industry developments,” she said.

One of my coworkers likes to call it ‘H.I.: human intelligence.’ ... And I think that the best practice that I’ve understood is the human intelligence can be on the front and the back end of it, and let the computer be in the middle.

Jane Clark, NationaLease senior vice president of operations

Image

NationaLease’s 107 independent truck rental and leasing companies are at different stages in their AI journeys, said Jane Clark, senior vice president of operations.

Clark said AI and machine learning will help fleets schedule maintenance at optimal times and avoid breakdowns. Systems will predict that a part should be replaced more accurately than relying on miles traveled. It will help shops avoid overstocks or shortages. Technicians can be more optimally scheduled because they won’t be responding to so many emergencies. Fleets will be better able to plan financially.

The Human Factor

Clark also noted there are challenges associated with adopting AI. Some people fear AI will replace people. She said those fears are largely unfounded. AI can’t fix a truck; instead, it can help people do it. Telematics and diagnostic tools coming from diverse vendors need to be integrated seamlessly. Clark suggested fleets incorporate pilot programs or gradually implement the technology to help staff adjust.

Clark said humans must give good prompts and input, and then check the work at the end.

“One of my coworkers likes to call it ‘H.I.: human intelligence,’ ” she said. “Not just artificial intelligence. We need human intelligence, too. And I think that the best practice that I’ve understood is the human intelligence can be on the front and the back end of it, and let the computer be in the middle.”

Image

Galland��

Another use for AI is OCR, or optical character recognition, which recognizes text in a digital image. If OCR reads a bill of lading with 90% accuracy, AI can flag the incorrect part and move it into manual review, said Hans Galland, president of BeyondTrucks, whose products help fleets streamline operational workflows.

Those kinds of activities allow humans to focus their mental capacity away from trivial matters and toward areas that are more complicated or require a human touch.

Galland said companies shouldn’t seek to use AI in some optimization decisions — for example, when there is a high cost of error and when AI can produce errors. One example would be loading white wine into an uncleaned trailer that previously hauled red wine. While AI usually has probabilities attached to it, that circumstance is more black and white.

“Essentially, anything that’s a lot of liability, you certainly would want to have a human in the loop to at least check and verify,” he said.

Reliable Data

Brian Mulshine is senior director of TMT product management for Trimble, whose products include asset maintenance software. He said data from past repairs can help fleets predict future events. But the data must be reliable.

“You can’t do AI on garbage data,” he said. “You’ve got to make sure you have good data structure. Start with good structure. By doing so, you can utilize your valuable repair history because you spent millions for it, because you’re paying on average a couple thousand dollars every service event.”

Users, Mulshine added, should take those service events and make them data friendly to better identify the trends and begin modeling the data for better predictions.

You can't do AI on garbage data.

Brian Mulshine, senior director of TMT product management for Trimble

Image

Using data and the right processes can positively impact a fleet’s maintenance operating budget and scheduling, asset utilization and uptime, Mulshine said. He added that fleets should collaborate with manufacturers to utilize AI most effectively.

“In the old days, I didn’t want to give any data to the manufacturer because they’ll probably void my warranty,” he said. “Things are changing today. The vehicles are connected. They possess better data, and it could lead you to better decision-making, particularly regarding warranty coverage.”

Adam Kahn, Netradyne’s chief business development officer, said better drivers can save 3 to 5 cents per mile in maintenance costs. Driver actions like harsh braking are a key indicator for brakes and tires. AI can recognize patterns and plot communications paths to the driver and maintenance manager, who can later recommend policy changes.

“Instead of using the term ‘AI,’ a lot of times what I’ll use is the term, ‘Act now. Talk later,’ ” Kahn said. “The information gets processed very quickly. You can generate some type of action that can correct the problem, and then later on, when the driver gets there, you can have a longer conversation.”

Image

Welbourne��

Evan Welbourne, head of AI and data for Samsara, said the technology can detect subtle shifts in operating patterns that can be early indicators of problems. If a trend in battery voltage historically has preceded failure, the system can flag it, and maintenance teams can proactively schedule actions.

Even with those capabilities, human oversight is needed to provide context that AI might not capture, such as a driver noticing a “strange noise,” Welbourne said.

Aman Singh is co-founder and head of analytics at Intangles, which provides an AI-powered predictive vehicle health platform. He said the AI knows what the driver doesn’t — that the engine temperature is slightly above normal — and will alert operations. Instead of a five-minute warning, the fleet learns what’s happening two or three hours before failure.

Singh noted AI is ushering the future of maintenance, and truck electronics will grow more complex because of changing emission standards. In short, AI is able to crunch more data than human engineers can.

AI can give organizations an outlook of “continuous optimization,” said Lytx’s Jeff Martin, vice president of global fleet sales strategy. Regressive models can see when parts have failed in the past and predict when things might fail again.

“This regression analysis allows for, if A+B, then C’s going to happen, and it’s just that simple,” he said. “And with all the billions of data points that are being collected annually, it’s only going to get better and more accurate.”

Abhishek Gupta, vice president of product at Motive, said fleets using AI and AI-powered image and video analysis experience 15% to 25% lower maintenance costs through predictive maintenance savings alone, as well as a reduced need for roadside repairs. The company has a 400-plus member safety team that reviews AI-generated data to filter inaccuracies and ensure it produces only relevant, actionable insights.

He said fleets must find vendors that can help them deal with challenges, such as integration with legacy systems, skepticism about accuracy and reliability, and concerns about driver and team buy-in.

“Transparent insights and a human-in-the-loop approach builds confidence in AI security and decision-making,” he said. “Fleet managers and mechanics may be hesitant to trust AI recommendations, fearing false positives or unnecessary maintenance. Accuracy is the difference between drivers that praise the tech and ones that want to throw it out the window.”

Want more news? Listen to today's daily briefing below��or go here for more info: