One benefit to AI is that it never becomes sleepy or distracted, and is always aware of its surroundings. Alerts can be sent to the driver to help prevent collisions if potential danger is detected. - Image: Gerd Altmann via Pixabay

One benefit to AI is that it never becomes sleepy or distracted, and is always aware of its surroundings. Alerts can be sent to the driver to help prevent collisions if potential danger is detected.

Image: Gerd Altmann via Pixabay

Last year, the U.S. Department of Transportation’s National Highway Traffic Safety Administration shared what was termed as “a crisis” and “sobering” data on the estimated motor vehicle crashes in the first half of 2021. Up 18.4% over 2020, the numbers speak to an urgent need to address roadway collisions and a grim reminder that the people who died are family members, friends, and loved ones to many more thousands of people.

April is Distracted Driving Awareness Month, which serves as a reminder that unsafe roads are a public health problem, and that distracted driving has become the number one offender. In the U.S., a pedestrian is killed every two hours and injured every seven minutes due to distracted driving, contributing to more than 3,000 deaths each year from this leading cause of roadway fatalities.

​​According to the NTHSA, driver inattention contributes to 93% of collisions. So what does this mean for the human driver – are we a lost cause in terms of improving our driving performance with mobile devices and interactive dashboards within easy reach while driving?

There is a best of both worlds scenario playing out today, particularly in the fleet industry, where AI is reinforcing safer driving behavior and advancing the National Safety Council’s Road to Zero initiative for all road users.

Human drivers can be extremely good. Consider the risk that a professional fleet driver faces with countless hours driving on all types of roads and in all kinds of weather conditions. A fleet driver’s risk of being involved in a collision is exponentially higher and yet many drivers have 40-year careers without a single collision – far better than the average human driver and far better than any autonomous vehicle on the road today.

Humans also excel at making good judgements in complex driving situations: multiple pedestrians at crossways, another vehicle that has to respond to an object on the road and having to anticipate what it and other nearby vehicles need to do to avoid it. However, AI is good at some things that humans are not. Nobody likes driving for hours on a boring road – we all become distracted or sleepy, but AI never does.

Today, computer vision technology and AI can look at a vehicle’s surroundings and understand external driving context such as nearby vehicles, signs, road surface, traffic lights, and pedestrian activity and fuse that information in real-time with the activity and behavior of a human driver inside the vehicle such as steering and breaking patterns, drowsiness, and diverted attention due to food consumption or a cell phone. The AI can make a decision as to whether the situation is headed in a bad direction or if there is an imminent collision within the next 5-10 seconds. If so, the AI will give the driver a warning and feedback and thereby help the driver take action to avoid a collision.

Today, AI is improving U.S. fleet safety by taking this idea and scaling it up to billions of miles and millions of vehicles for continual improvement. AI-generated audio feedback through alarms and voice commands is giving thousands of commercial fleet drivers critical extra time to take corrective action. That same feedback also acts as a gentle nudge that measurably and automatically improves driver performance and reduces risky behaviors without any intervention by a human coach.

Consider that it takes about 1.5 seconds to react to an accident on the road, and hitting the brakes requires another 1.5 seconds. Depending on speed and road conditions, a typical driver needs at least 3 seconds of advanced warning in order to help prevent a collision. These extra seconds of advance warning are critical, however most traditional telematics and dashcam solutions fall short on accident prevention. Legacy technology often depends on video upload to the cloud and back again, such that by the time the data is transmitted, any promise of a real-time warning has passed

The path to solving the monumental challenge of distracted driving is correlated to giving drivers accurate warnings with as much extra time to react as possible. The way to achieve this level of precision is through predictive-AI technology that resides on the vehicle so that it can reliably serve as a trusted co-pilot for all drivers. Further it is important that predictive AI technology is able to accurately detect as broad a range of risk scenarios as possible; extending the most common lead vehicle use case to also include pedestrians, bicyclists, and motorcycles no matter whether operating day or night, in all weather conditions, and in any road context.

Fortunately, Congress is taking meaningful steps to combat the problem of distracted driving with the recently passed infrastructure legislation. It contains a key vehicle safety provision that includes the study of collision warning technology. The Stay Aware for Everyone (SAFE) Act requires the Department of Transportation (DoT) to study how driver-monitoring systems can prevent driver distraction which caused more collisions and fatalities on U.S. roads in 2020 than in any previous year, despite the fact that we were in the midst of a pandemic where drivers were on the road less.

About the Author: Yoav Banin leads product strategy and design for Nauto’s driver and fleet safety offerings. Prior to Nauto, he was Co-Founder and CEO of Solergy, a solar energy technology innovator and manufacturer. Yoav has also served as Strategic Advisor to venture-backed, enterprise AI startups which raised follow-on rounds or were acquired. Previously he held product leadership positions at Mercury Interactive (acquired by HP). Yoav holds a B.A. in Applied Mathematics from UC Berkeley and an M.S. in Materials Science from Stanford University.

Originally posted on Automotive Fleet

0 Comments