The role of telematics in fleet operations is expanding rapidly and nowhere more so than for LCV fleets. As new AI technology develops, fleet operators will only get more data to enhance safety, as John Kendall finds out…
Many older drivers were attracted to van and truck driving because once they were in the cab and on the road, they could go about the job in the way they chose. As long as the work was done satisfactorily and on time, fleet manager and driver were happy.
As costs have risen – the seemingly onward march of the price of fuel, as well as the rising price of vehicles and greater legislative requirements surrounding vehicles and drivers – fleets have inevitably sought out ways to contain those costs. Telematics was a solution that many fleets were drawn to, initially as a means of reducing expenditure.
Although an added cost in itself, telematics has shown that the potential savings can far outweigh the costs. Since costs don’t stand still, smarter systems have come along, initially camera-based to provide added information surrounding incidents, but now enhanced with machine learning and artificial intelligence (AI).
As several people that we have spoken to have said, AI and machine learning have become buzzwords, so it’s worth explaining what they are and how they can be used. Effectively it is developing software so that it can recognise certain patterns. Much of the focus is on driver behaviour, but there is much more that the systems can do besides.
James Dewhurst, corporate sales manager for the UK and Ireland at Webfleet, gives an example: “We have a tyre pressure monitoring system. It takes a couple of factors in, it takes the temperature and it measures the pressure and warns if a tyre is going to blow, so we’re stopping a potential blowout. AI can warn the driver and tell him he needs to pull over.”
Craig Allan, director strategic partnerships at ABAX suggests another. “If you are taking on a new delivery contract, AI/Machine Learning will be able to quickly review all the routes required to be taken to service that contract and cross reference this with the historical safety records of that route. This will provide your fleet with the statistically safest routes and times, cutting down on accidents, keeping your employees safe and your fleet downtime to a minimum.”
Damian Penney, vice president EMEA at Lytx, sums up what most fleets will see as the principal benefits. “By providing a proactive approach to fleet safety, MV+AI technology is empowering drivers and managers to mitigate risks before they become incidents. Drivers don’t have to wait until they get back to the depot to be told they had a momentary slip-up where risk increased. Instead, the power is put in their hands to make instant changes to driving behaviours. As well as keeping drivers safe, this FNOR (First Notification of Risk) -based approach is also helping fleet managers to reduce insurance costs. Insurers can get an accurate risk profile of the fleets they cover, helping them to make smart, data-backed decisions when calculating risk and adjust premiums accordingly.“
Penney refers to MV – Machine Vision rather than Machine Learning – but it is effectively machine learning applied to camera-based systems, where cameras can be enhanced by algorithms that can recognise specific things that the camera records.
Picking up on his point about insurance, Craig Allan at ABAX makes an interesting observation: “Currently when you hire a driver, you know their age, you can access the DVLA records and that’s about it. Imagine if you could, using AI, look at their entire driving history and use this to obtain a better insurance premium.
“AI/Machine learning can take a full driving history – where, when, how and provide an individual score that is much more accurate than date of birth and penalty points, meaning that your fleet has a much more accurate (and hopefully cheaper) premium.”
While telematics systems can be an important tool for helping to manage risk, Barney Goffer, UK product manager at Teletrac Navman, underlines how AI can also help fleets to identify good behaviour on the road too. “As a manager, when you’re debriefing somebody or you’re doing an analysis on risk within the business, we’re focused on the things that really are the riskiest. It’s using AI technology, it’s using machine learning as well so the camera itself is constantly working out what good looks like and what bad looks like on driving conditions.
“And as a manager you can reject events as well. So if something came through that was deemed severe, say it’s a speeding event for example, let’s say the road speed data was incorrect on that particular road and you knew that as a fleet manager. You can reject that event, then next time, the system learns that and it’s got the ability to recognise road speed signs, traffic lights, all these different things and it builds up a really realistic picture of what that individual is like in the cab, be it good or bad, but in particular the good stuff and that’s really the focus for us.”
Fleet managers using their judgement about what is most important was an issue raised by Steve Thomas, managing director of Inseego, until recently known as Ctrack. “If you’ve got a large fleet of several hundred vehicles you’re going to have a lot of speeding incidents. The vast majority will be 1, 2, 3mph over the limit and whilst we’re not condoning that, it’s not as risky as someone doing 40mph in a 20mph zone past a school at 3.15pm in the afternoon, when there are hundreds of kids lining the streets, all potentially looking at their mobile phones, all potentially about to step into the road without looking. It’s allowing our customers to see where the real issues are first. If you’ve got 200 drivers and they’re all speeding, which one are you going to talk to first? It’s the ones that are speeding the most in the highest risk environment and that’s what we’re helping people to do.”
Geotab has been using AI to help identify various vehicle-based matters. This includes assigning the work that a van is used for to each vehicle, so that meaningful comparisons can be made between vehicles on similar duty cycles. Aaron Jarvis, associate vice president sales and business development at Geotab, discussed how the company is using AI to help fleets transitioning to electric vehicles. “Some companies have asked, ‘Where are our vehicles parked overnight?’, because that’s where they want to put their charging infrastructure. Lots of customers ask that so we build it as a tool into the portal, so you can click on it and see where your vehicles are parked.
“Another question we get asked is, ‘What’s the average idling for a similar fleet?’ We’ve built that into the portal based on customers’ requests so it’s important and it’s important to implement it in the right way – to either try and sell people something or you can build something that adds value.”
Using AI to bring in data from other fleets is also an advantage that James Dewhurst at Webfleet raises: “You might have a list of all the harsh braking and harsh steering events. What you don’t have is what the weather was at the time of that event, or the road conditions. We can then bring in additional data and we’re already doing this with some customers and maybe also looking at the type of road they’re driving on – a motorway, urban roads?
“We did this exercise for a customer’s fleet in the Netherlands. We looked at their data and we added in these extra algorithms and then came up with a more accurate risk-based factor and we ran 16 different machine learning models to make sure it was correct.
“We were able to show them that their fleet was twice as risky as other people in the industry because we took the data from all the other fleets we had in their industry and also all the vehicles we had in the Netherlands and benchmarked it against them. To do that manually would take thousands and thousands of working hours,” he concludes.
Richard Lane, commercial director at VisionTrack, illustrates the large numbers that fleets can be faced with once you start to consider the data that telematics systems can deliver. “Just to think of one particular customer, they’ve probably got around 850 vehicles on our system, covering roughly 1.2 million miles a week. That’s 45,000 hours of driving, or 47,000 trips roughly in that time. If we just look at the raw data coming out of the device – where it has breached a driver behaviour threshold, that has happened around 170,000 -180,000 times in a week.”
VisionTrack has used AI to develop a system that will analyse video clips. Again, while it may concern driver behaviour, part of the intention is to make rapid decisions about what needs to be done in the event of an incident.
In those circumstances, AI can be used to take action that could save someone’s life, as Lane outlines: “A computer vision service that operates 24/7, 365 days a year will effectively, within a matter of eight seconds make a determination on a crash. This could include whether or not the driver needs urgent medical assistance or to start a First Notification of Loss (FNOL) process. That is a real game changer in terms of the utilisation of the machine learning and computer vision technology.”
