How AI-powered coaching helps drivers, trucks, and AI itself

“AI is definitely useful for coaching, especially since the most useful coaching is done by AI in real time as you’re driving and focusing on the positive/correct response,” said Dr. Stefan Heck, CEO and founder of Nauto. “It shows what you should do compared to what you’re doing, and for Nauto, we’ve seen that 85% of drivers do change their behavior on their own without any manager intervention as a result.”

And that change in behavior can then increase vehicle lifespans by decreasing wear and tear on components. For example, Rajesh Rudrarahya, CTO at Lytx, noted that feedback from AI can help limit aggressive braking by notifying drivers if they’re following too close. This helps the driver avoid suddenly hitting the brakes and the associated wear that comes with that behavior.

“If you think about the wear and tear impact of a lot of these things, they cause maintenance cycles to happen sooner than later,” Rudrarahya noted.

He noted a similar phenomenon on fuel efficiency, where AI can help coach drivers to avoid idling for too long. Instead, AI coaches can advise drivers to, perhaps, take their break elsewhere instead of in their truck, and thus decrease wear on a truck’s engine or emissions system.

But knowing a technology has potential and actually being able to get concrete benefits are two different tasks. To do this, Heck said fleets need to be able to establish economic reference points of the technology’s impact, not to mention figure out how to implement AI technology effectively. To help, here’s how both Lytx and Nauto offer AI options for driver coaching.

AI coaching for drivers

For AI offerings, Lytx provides an AI-powered dashcam that can also provide light and audio alerts when it detects drivers braking hard, swerving, or using their cellphones. Meanwhile, Nauto’s safety platform can also provide driver alerts when it detects driver drowsiness, tailgating, or speeding, and provide audio alerts to help a driver self-correct.

While these features alone can help drivers be safer on the roads, when used in tandem with a human coach (in a method Lytx calls ‘blended coaching’), their benefits can compound. For it to work, the company advises fleets combine their traditional video- and document-based driver training with AI-powered alerts and in-person coaching. This way, drivers can learn what’s expected of them, get a reminder of those expectations from AI while they’re in the cab, and then their manager can follow up with them on why that reminder was important and answer any questions or concerns out of the cab.

This works better than if a driver only had that original training to rely on, Rudrarahya said, because Lytx has found only 20-30% of drivers tend to remember training materials unless it’s reinforced. And it improves upon traditional coaching methods where a manager might coach a driver on an incident from weeks prior, at which point “the conversation is not effective because of the poor timing [of their feedback].”

“In the blended coaching approach, what does happen is you're talking to drivers about individual events that a camera has captured with AI, then there is this conversation on what you could have done better in this specific situation,” Rudrarahya explained. “It builds that reinforcement loop in their head on what leads to an incident and how they can overcome that.”

Benefits for AI technology

Plus, not only can AI-powered coaching help improve drivers in the moment, but the drivers can help improve the technology, too. For example, Heck noted that Nauto also tracks how drivers respond to their AI alerts to make them more effective.

“Did the driver respond to the alert?” Heck asked. “Did the driver give us a thumbs up or did we get a middle finger? We use that to retrain the AI to make sure we're intervening, not only accurately, which everybody knows is important, but also with appropriate timing.”

This helps AI offerings warn drivers more effectively in the future, without providing warnings so early that a driver ignores the technology, or too late so that it becomes redundant.

After all, “Nobody wants to believe a black box that's telling you to be alert or stop speeding,” Lytx’s Rudrarahya noted. “That's a common pitfall in the beginning.”

That’s why this driver-AI feedback loop is just as beneficial for the technology as it is for the driver.

“It's really important that when we intervene, the driver actually recognizes, ‘Oh, sh--, that was dangerous,’ and that's when they learn and change their behavior,” Heck concluded.