Trucks
Trucks
Modern-day trucks generate vast amounts of data every minute they’re in operation. But how is this data being used? How can it benefit truck owners? And what does it mean for the future of trucking?
Today, a typical heavy-duty truck is fitted with over 100 sensors. A smartphone, by comparison, has around ten. Every minute, a truck will send around 20 gigabytes of data, which is the equivalent of streaming 1800 hours of music. In that same minute, we will receive over 600,000 different metrics and three million log messages.
Now, multiply that one minute by the number of minutes in a truck’s operational life, and the amount of data being generated becomes inconceivably vast. Yet, far from drowning in all this data, the data scientists working in the industry crave even more.
“The more data, the better,” explains Robert Valton, Head of Data, Analytics & AI, Volvo Group. “With our data science competence and the advanced analytical methods and tools we have at our disposal, huge amounts of data are not a problem – it is an opportunity. It enables us to generate even deeper insights into the truck’s behavior and better understand how it works to optimize the transportation and the support to our customers”.
In the early 1990s, the first connected trucks were launched, and the number of connected vehicles has seen linear growth ever since. The amount of data being generated has grown exponentially over the past 30 years, but the challenge has been to find ways of using this data to create value for truck owners and transport businesses.
“The history of how we use data from trucks can be viewed in four phases,” says Robert. “First, we were reactive and looked at the data to determine: what happened? Then, with connectivity we started looking at data more in real time and determining: what is happening? In recent years, we have been addressing what will happen and taking action to prevent it – real time monitoring is a good example. Now, we are going even further and using data and AI as a crystal ball to determine what we would like to happen to best support our customers.”
A truck’s data can be used to predict and prevent breakdowns, and by extension, improve uptime.
Connectivity lies at the core of preventive maintenance – the concept of predicting and preventing breakdowns before they happen.
By analyzing the vast amounts of data that can be extracted from vehicles, and applying machine learning, it is possible to identify common patterns and combinations of factors that lead to a specific fault. This can then be used to create models for predicting and preventing similar faults in other vehicles.
“We send the responsible workshop an alert so that they can schedule a convenient time for the customer to visit and diagnose the issue before it results in an unplanned breakdown,” says Elke Decaluwé, Vice President, Technical Dealer Support, Volvo Trucks.“ For customers, this means increased uptime and avoiding all the costs associated with a breakdown, such as loss of income and damage to the company’s reputation.”
Today, Elke and her colleagues collect data from a fleet of nearly 85,000 trucks operating across Europe. Their work has changed dramatically in recent years with new advances in connectivity and data analytics.
When they started in 2016, they were monitoring a fleet of just 600 trucks, for one component – the battery – and it took a full day to complete one check. Now, 11 different components are monitored and a check can be completed every eight minutes. Around 4,000 alerts are sent out each month, of which it is estimated that 77% prevent an unplanned breakdown.
However, with the pace of development not slowing down, the data models and algorithms continuously need to be refined and improved.
“Trucks are not static and are constantly evolving, so the data is evolving too,” says Elke. “If we miss a breakdown, or an alert does not work, then that’s a trigger to take a closer look and see if our models need to be tweaked.”
“With AI, we can conduct even more of the analysis onboard the truck itself...It would almost be like a cognitive and self-healing truck”
The evolution of AI has the potential to make the current models even more accurate and comprehensive. Since AI has the capacity to analyze far greater amounts of data, it can identify previously unseen and unknown patterns and connections between data points.
“Traditionally with data analytics, you take a hypothesis-driven approach where you select the parameters you believe are relevant,” explains Robert. “With an AI-driven approach, you look at all the available data from the truck, regardless of whether you think it’s relevant. We can also blend in other data sources, such as weather and transport conditions. We can create models that are even more accurate and can look further ahead.”
AI could also pave the way for even smarter trucks – vehicles capable of effectively diagnosing and repairing themselves.
“Today, we send data from the truck to a monitoring center’s backend. But with AI, we could conduct more of the analysis onboard the truck itself. If it encountered a problem, it would automatically run the diagnostics and resolve the issue via software changes. It would almost be like a cognitive and self-healing truck that can optimize uptime and enable more transport with less climate impact,” says Robert.
