The use of artificial intelligence (AI) in today’s fleets to monitor and analyze data is growing. “AI is a powerful tool. When paired with machine vision (MV) technology in a video telematics solution, AI can provide fleets with the valuable data and insights needed to enable safer and more efficient fleets,” said Michael Philippi, vice president of technology for Lytx.

There are numerous benefits for an AI-based system beyond traditional telematics solutions. Nine added benefits of utilizing an AI-based telematics system can include: 

1. Reduce Human Error 

Likely the top benefit of an AI-based system is the possibility of reduced human error.

“Using machine learning and computer vision, we trained an algorithm to do video classification or identification, rather than having human beings manually review videos. Our program is designed to detect patterns in large data sets and to predict outcomes with reliable, effective, and repeatable results, and helps filter different Onboard Event Recordings (OER)-triggered events by categorizing videos as primary or secondary risks. This brings the most critical videos to the forefront so fleets can take immediate action,” said Chris Orban, vice president of data science for Trimble Transportation.

There are also advantages of using AI in terms of speed. 

“We can process a video through our system within minutes, rather than possibly hours or days with a human-based system,” Orban added. 

2. Increase Objectivity

Another top benefit of using AI with a traditional camera system is that the system is entirely objective. 

“Every time a certain kind of event happens, you will get the same answer back from an algorithm, as opposed to a human who might see one situation, a certain way one day, or another way the next day,” said Orban of Trimble. “To be clear, I’m not saying that the human being is necessarily wrong, and the computer is right. But the computer does the same thing repeatedly. If the computer is making a mistake, we can identify and correct it. It’s much more difficult to get a human being to change how they look at a hard-braking scenario.”

3. Hold Drivers Accountable

One step up in sophistication can detect courteous driving and award drivers points that can incentivize better habits. 

“AI-equipped cameras can also turn good driving into learning opportunities for other drivers. And, fleet managers can also reward safe drivers with the data,” said Ryan Driscoll, VP of marketing for GPS Insight.

The J. J. Keller solution uses AI to read and interpret posted speed signs. 

“The benefit of this is that the driver is held to the actual speed limit instead of the one that may be stored with the route, interstate, or road, which can often be different than the actual speed limit, due to construction, etc.,” said Mark Schedler, Sr. Editor – Transport Management for J. J. Keller & Associates, Inc.

4. Analyze in Real-Time

One of the most powerful benefits of AI technology within telematics is its ability to provide real-time feedback to drivers, to help avoid accidents before they occur. 

“These in-cab alerts enabled by AI are designed to act as an ‘extra set of eyes’ for drivers. AI capabilities within your telematics platform can also streamline the analysis of safety-related data - such as distracted driver behavior, tailgating, near collisions, and more. This eliminates guesswork and enables safety managers to operate their programs even more efficiently,” said Ingo Wiegand, director of Product Management at Samsara.  

5. Detection without G-Forces

Some systems require G-Forces to trigger events. While this tracks braking, acceleration, and turning, there are several other distracting behaviors that can lead to an incident. 

“Earlier this year, Lytx released four new AI triggers focused on distracted driving and other high-risk behaviors that occur inside the vehicle: cellphone use, food or drink, not wearing a seatbelt, and smoking. Fleets can now identify these risk or distracted behaviors independent of another trigger or event and help fleets’ address them before they result in a collision,” said Philippi of Lytx.

6. Track Trends

AI algorithms highlight key trends that can be used to inform fleet management better. 

“For example, a dashboard allows maintenance professionals to see vehicles with recurring issues, operations, and sub-units with higher numbers of engineering-related events and which vehicles should be prioritized for maintenance,” said Jonathan Bates, Head of Global Marketing for MiX Telematics.

7. Ability to Automate

Automated safety features include automatic braking, lane departure systems, and proactive driver coaching. 

“Automation helps reduce risky driving behaviors before they become driving errors, such as hard braking, harsh cornering, and speeding,” said Sherry Calkins, Vice President Strategic Partners for Geotab.

Other common driving errors automation sets out to correct are idling and “phantom” traffic. 

“Idling results in the emission of excess CO2 which is harmful to the environment. Automation features allow fleet managers to coach drivers to help reduce idling and curb their fleet’s overall CO2 emissions,” Calkins explained. “Phantom traffic is caused when a single driver breaks suddenly and unexpectedly, causing a wave of drivers behind them to break, causing delays on the road for no reason. Automation would help alleviate this issue because automated vehicles could brake, and accelerate in unison, potentially preventing this type of traffic entirely.”

8. Maintenance Visibility 

When you build AI into the dashboard, you can also identify vehicle makes and models that may require attention in the future.  

“The software shows where the problem areas could be in the future without engineering teams having to review multiple reports to detect possible fleet engineering issues. Maintenance professionals can optimize the use of repair garage time, make sure that vehicles are serviced before an event becomes a problem, and spot recurring issues across multiple vehicles of a certain type,” said Bates of MiX Telematics.

9. Improve Driver Health & Safety

AI is also used to improve fleet safety by analyzing data. 

“A machine learning model can learn from this data to help identify driver risks such as distracted driving. Furthermore, AI can go a step further by predicting driver behavior and driver efficiency, which can help businesses improve their services,” said Calkins of Geotab.

UPS is one of many of businesses that has employed AI to improve driver health and safety. 

“One of the main focuses of UPS’s efforts was to improve the safety of backing events. By utilizing AI and focusing its efforts on proactive fleet safety, UPS was able to eliminate driver backing patterns and gained a driver safety commitment to avoid backing up in specific situation,” Calkins said.

10. Data Mining for Improvement

The foundation for AI-based systems is data. 

“Data can provide actionable insight that can improve operations, safety, and process saving fleet managers time and money. Setting up a system now will give fleet managers a competitive advantage by giving them increased visibility of their operations that can open up the doors to digitally transforming their operation for the long run, keeping their drivers safe, minimizing expenses and ultimately, securing their role in the evolution of fleet operations,” said Jeff Clark, senior vice president of Product Management for CalAmp.

Clearing Up AI Misconceptions

The benefits of an AI system are clear, but as with all newer technology solutions, there are several misconceptions about the system and its use. 

1. All AI is not the same. 

It’s important to understand that, just because a program is built with AI, doesn’t mean it’s necessarily smart. 

“You need to be able to assess the effectiveness of the AI and apply it to the problems it’s uniquely good at. Don’t put AI in a situation where, for instance, it’s beeping but not doing anything to correct the problem it was created to correct. AI has to be in the service of the whole, rather than just a single stand-alone piece of technology,” said Ray Ghanbari, CTO for SmartDrive.

On that same wavelength, not all dashcams are the same. Features range from basic forward-facing cameras that store a few gigabytes of data locally to AI-powered cameras with 360-degree views. AI-equipped models can help you coach drivers.

Fleets should consider the credentials of an AI company that is delivering a solution. 

“For example, there are companies that license their algorithms… If they don’t own their own software stack, they could be slow in updating the system/technology,” said Adam Kahn, president of Netradyne. “AI gets better and better as it learns just like humans and is incredibly accurate.

2. AI and privacy infringement concerns. 

There are sometimes questions about AI and its role monitoring and how it might infringe on privacy. 

“Solutions need to be evaluated with privacy concerns in mind. Ultimately, AI is aiding the work of drivers and managers to increase productivity. For our dashcam solution, the AI system is triggered only by a harsh driving event and is a road facing dashcam. Even when it records an event and flags it, fleet managers or administrators still determine if the categorization is correct. In this way, the solution is helping to increase the response time of fleet managers and protecting drivers if they need video evidence in the event of an accident,” said Kevin Aries, head of global product success for Verizon Connect.

3. AI doesn’t solve every problem in the world. 

AI will help take organizations to the next level and, if used properly, will make them more efficient. 

“However, the most important building blocks of a safety program must be put in place by human beings, dealing with other human beings. The key change levers are the human interactions. AI will provide better insight into what should be done, but humans must communicate effectively to do it,” said Bates of MiX Telematics.

4. AI isn’t what we see in science fiction. 

AI is not a machine sitting in the background doing all the thinking and making decisions. 

“Computers evaluate the information they are given based on the parameters that we provide to them. So, we must teach the computer something before it can have any intelligence to help us make decisions,” said Orban of Trimble. “The key advantage is the speed of the computer and the ability to analyze many, many, different factors at the same time.” 

Orban noted that it’s not that the computer is smarter than a person -- the computer must be trained to understand the situation that it is being asked to evaluate, so you still need humans to do that. 

“We’re not going to remove humans from the equation. Instead, we’re trying to let the computers do what computers are good at and let humans do what we are good at,” Orban added.

Why is Machine Learning Important?

In a nutshell, artificial intelligence (AI) or machine learning is also used to improve fleet safety by collecting data. Machine learning is a subset of AI. 

“Machine learning automates the process of analytical model building. We’re often asked if machine learning is the same as predictive analytics. When deployed for business and safety purposes, which is what we do in telematics, machine learning is predictive analytics. In a nutshell, machine learning means the computer learns without being programmed to do so,” said Reza Hemmati, VP of product management for Spireon.

Machine learning is a core element of Spireon’s IntelliScan cargo sensing technology, which uses optical imaging and laser time-of-flight to give accurate trailer cargo readings available.

“Our system uses machine learning to get better at what it reads as it is used. Whether you are hauling standard cargo in crates or on pallets, irregular cargo such as canoes, or soft cargo such as paper or carpet, IntelliScan’s machine learning technology adapts, learns, and incorporates its findings into new algorithms to make it nearly infallible,” Hemmati added.

An important component is the learning system itself. 

“If a program is static, it’s hard to make improvements. What you want is a program that is continually updating what you measure to optimize the safety ‘needles in the haystack.’ As driver behavior changes, you want your safety program to automatically adapt to the next safety issues you want to focus on. This is the backbone of our advanced analytics efforts and are even more critical during these times. A static safety program that you may have optimized or tuned, a few years ago may need an adjustment today. Our adaptive learning AI systems are, in real-time, continuously adapting to what’s happening on the road—not just at a customer level—but being informed by the collective experience of all our customers and what they see across their fleets,” said Ghanbari of SmartDrive.

Verizon Connect’s Integrated Video dashcam uses AI and machine learning to help businesses improve fleet safety and protect the business. 

“When a harsh driving event occurs, Integrated Video captures a clip of the event and using AI, classifies it by degree of severity, and alerts the fleet manager. Through machine learning, Integrated Video learns and gets smarter about what the business considers harsh driving to improve the accuracy of future clips. This helps fleet managers spend less time reviewing clips and more time proactively addressing accidents or harsh driving behavior when it occurs,” said Aries of Verizon Connect.

The Bottom Line

Systems are only as good as the work that you put into them. 

“There’s no system that I’ve seen where you can just apply a true artificial intelligence to your business and wipe your hands and say, ‘I’ve solved my problem.’ It always takes work upfront to make sure the inputs to a system are going to be clean and valuable. And, it takes work to make sure you’ve selected the right kinds of algorithms and to make sure the computers have access to the right tools to do their job,” said Orban of Trimble.