AI improves driver safety with actionable real-time insights

Crashes and unsafe driving incidents are ever-present concerns in the trucking industry. 

According to the Federal Motor Carrier Safety Administration (FMCSA), there were 4,447 fatal crashes and 114,000 injury crashes among large commercial vehicles in 2019. These were largely the result of distraction, inattentiveness and poor driver decisions, according to the U.S. trucking industry’s regulating government agency.

What’s more, from a business standpoint, the average cost of a crash is $91,000. That number increases to $200,000 if injuries are sustained, and rises as high as $3.6 million if crashes involve fatalities. 

When AI takes the wheel

A new study from a leading fleet management company indicates that AI technologies can go a long way toward curbing such tragic – and damaging and expensive – events. 

KeepTruckin today released the results of a two-year-long study on the use of AI technology for driver safety and coaching in commercial vehicles. The key findings: active use of AI and coaching techniques led to 22% fewer accidents and 56% fewer unsafe driving incidents. 

“We wanted to understand whether the use of our AI technology is reducing accidents,” said Abhishek Gupta, KeepTruckin’s director of product for safety, compliance and insurance. 

The study indicates that it has, Gupta said, and he underscored the fact that “proactively identifying and modifying high-risk behaviors with a robust safety program can help save lives and ultimately drive significant bottom-line improvements for businesses.”

KeepTruckin’s integrated fleet management technology is designed to leverage edge computing, computer vision and real-time insights and alerts to help improve safety, productivity and profitability for commercial fleets, Gupta said. Notably, its AI-powered dashcams continuously monitor drivers and alert them instantly – and via audible channels so as to not present their own undue safety risks – if unsafe driving behaviors are detected. These can include speeding, tailgating, unsafe lane changes, cell phone use, eating behind the wheel or improper (or no) seatbelt use. 

“We want drivers to be instantly alerted,” Gupta said. “This leads to preventative action.”

Cams also continuously record drivers, and collected data helps establish driver profiles. This is then provided to back-office fleet managers who can facilitate ongoing coaching efforts. Drivers can also receive automatic coaching through a KeepTruckin Driver App. 

“They can take this information and start coaching the driver,” Gupta said. “It is important for drivers to answer the question, ‘How should I be improving my behaviors?’”

Keeping the fleets moving and safe

The nine-year-old fleet management hardware and software technology provider has customers spanning from small operators to large fleets. These collectively have roughly 700,000 commercial vehicles moving from 12 to 15 hours a day on U.S. highways and byways, Gupta said. All told, KeepTruckin’s base comprises more than a million drivers.

Study analysis included more than 5,000 fleets operating at least 20 vehicles. The San Francisco-based KeepTruckin tracked several metrics over two years, including speeding and other unsafe driving behaviors such as close following, distraction, drowsiness, and seat belt violations, as well as what are known as “harsh driving events” involving abrupt braking, cornering and acceleration. This data was collected from both road-facing and driver-facing dashcams. 

The study also found that fleets that implemented KeepTruckin dashcams and consistently coached their drivers saw the following: 

• 69% fewer unsafe driving behaviors such as cell phone usage and tailgating
• 22% fewer FMCSA-reported crashes
• 40% fewer FMCSA-reported unsafe driving violations
• 67% fewer harsh driving events
• 31% fewer speeding events

Gupta pointed to KeepTruckin’s capability to pull from such a wide survey area, and over a considerable period of time. By contrast, other field studies have harvested data from much smaller driver pools and time periods – for example, just a few dozen drivers over a course of a few weeks. 

“From an analytical rigor standpoint, that’s really important,” Gupta said. “Nobody has this data across every road segment. We’ve been able to collect a wealth of really unique data and turn that data into actionable insights.”