Geotab says AI can reduce fleet failures and make roads safer

AI and fleet data have great potential to improve vehicle safety and reliability

Getty

AI has impacted every aspect of our lives, but it has become generative, especially now, and has dramatically impacted the automotive industry. Chinese manufacturer XPENG even believes that vehicles have an AI-defined future, and Volvo will help train safety models using an AI technology called “Gauss Splatting.” Some of these changes are not found in the car itself, but how its maintenance and road use is managed, especially the fleet. I spoke to Edward Kulperger, Senior Vice President of Geotab, a leading car telematics company, about how AI is improving reliability and making roads safer.

Geotab's long history with AI

“We've been digging into AI for over a decade,” says Kulperger, claiming that Geotab is ahead of the curve. “Neil Cowes founded an organization from his basement 25 years ago. We currently have 2,500 people and serve the world's largest companies. He has been investing in data programs, scientists and more since 2013.

“They were working with different fleets to help them move to electric vehicles at the time,” says Kulperger. “We have now developed several attractive electric vehicle suitability assessment tools and infrastructure assessment tools,” which can assess, for example, what is needed to move from internal combustion to electricity, from large last mile delivery companies. The tool supports migration and how the company has enough power to operate. The company also offers some of the most comprehensive EV battery life data, indicating that electric vehicles are likely to last much longer than some people expect.

AI is playing an increasingly important role in Geotab's technology-centric investment strategy. “AI godfather Geoffrey Hinton spoke at Geotab Connect, Geotab's annual conference in 2018,” says Kulperger. “He's just won a Nobel Prize in physics, but he's a professor at the University of Toronto. We were based just outside of Toronto and bridged the conference, where he gave one of the keynotes.

Geotab has been using AI technology for several years, and LLMS has been sending shockwaves through creative circles, but it has the problem of hallucination in which the generated AI invents information. This is a problem if you are using a search engine, but it can be very serious when you provide incorrect information to your car management system. “It's obviously incredibly tricky,” says Kulperger. “But we have 150 data scientists considering the data.” These data scientists make sure the output is correct. “We have a complete monitoring program within the data.”

Predicting failures using GeoTab AI

“We have developed other AI tools to see the last miles organization that is delivering in our community,” says Kulperger. “They start and stop the vehicle thousands of times a week. It puts incredible stress on the batteries, starters, alternators. We looked into this pattern. It takes time to start a vehicle to hit the starter from that electronic pulse. The starter, alternator, or battery breaks down, saves millions of dollars, get out and adopts it four years ago, pulling down fleets of all sizes.

This is just an example of how Geotab is harnessing the power of AI, and Kulperger believes it will speed up the pace of development. He believes that the public and private sectors are increasingly pumping data, processing power and energy into AI. “Things are accelerating exponentially, even what we thought,” he says. “Forecasting next year is extremely challenging.” GeoTab has developed a platform to promote clients who use data, including AI. The platform can be used for AI models built on top of innovative safety tools, Vision Telematics, and Geotab software.

This approach pays dividends when analyzing accidents, ensuring that they can be mitigated as cost-effectively as possible. “We have our own data scientists who focus on quickly extracting data so that we can quickly extract data to find out what happened in the accident and use neural networking to jump into the data lake to see if we've seen similar accidents in the past,” he says. “They can see what happened in an accident like speed, G-force, RPM, brakes, time of day, road conditions, etc. They can then rebuild an accident that will help customers save money from a repair perspective. They usually have a deeper understanding of what happened in that accident.

GeoTab makes it easier for clients to access data using generated AI. “We're building that foundation so that our partners can create simple queries, such as asking the energy they used in their vehicles in Birmingham or London last week,” says Kulperger. “That query goes out into the model and gives you an answer. It also provides a way to get that answer. That's important, but if you look beyond that and look at the data lake, you'll see around 80 billion data points every day. That's an astronomical amount of data.”

There will be more to come from applying AI to GeoTab's huge dataset. “There's a partner methodology, and obviously privacy and security are paramount,” says Kulperger. “But partners will be leveraging data lakes to be aggregated and anonymized to enable public organizations to improve their infrastructure, such as placing speed limit signs and knowing where to reduce accidents and dangerous hotspots that are happening.