Forget autonomy: Here's how driving becomes safer

Commercial fleets are an important bellwether for the future of safe driving.

nauto.jpg

Traffic accidents are a misnomer. In fact, over 95% of what happens on roadways is perfectly predictable. 

According to Stefan Heck, PhD, CEO of Nauto, a leader in AI-powered advanced driver assistance systems. With the belief that fully autonomous driving is still years away, Heck's answer to the mounting number of collisions and fatalities in 2020 is to help drivers, not replace them. To that end, Nauto's technology underpins sophisticated safety systems for hundreds of the world's top large-scale fleets, and customers are achieving up to an 80% reduction in a collision loss. The company estimates that has translated into over $300 million in savings.

see also

Artificial intelligence in the real world: What can it actually do?

What are the limits of AI? And how do you go from managing data points to injecting AI in the enterprise?

Read More

So how can fleets operate safer and more predictably? And what does that mean for non-commercial drivers and pedestrians? I caught up with Heck, who shared interesting insights on the very human future of driving.

GN: A lot of attention is paid to autonomous driving, but a fully autonomous future is still years away. How does Nauto's technology augment driver awareness and safety with AI?

Stefan Heck: Nauto gets 90% of the safety for fleet and risk reduction benefit of autonomy today for 1/100th of the cost. AVs made rapid progress on the first 80% of driving - staying in lanes, avoiding other vehicles but the last set of conditions of AV driving are really hard - left turns, crowded urban areas, complex interactions with other road users. 

Nauto's AI can also operate in any environment under any weather conditions and any type of vehicle well beyond the range of conditions that AVs can drive in safety today. We basically provide a "co-pilot" for commercial drivers who never falls asleep or gets distracted, helps them get to their job site and home safely, and embodies the collective experience of billions of miles of understanding risk and safe behavior available to every person driver at a moment's notice.  

GN: When it comes to commercial fleets, how do accidents tend to occur and how does your technology mitigate collisions?

Stefan Heck: By far, the biggest cause of collisions in commercial fleets (or, in fact, any vehicle) are distractions. We see about 70% of all collisions happen to drivers that are frequently distracted. Fleets with long distances or hours also see large numbers of collisions due to fatigue and drowsiness, often at the end of a long grueling day of physical labor -- e.g. installations, package delivery. Speeding gets a lot of attention but is actually mostly an amplifier of damage, not the primary cause of collisions. Similarly, many older commercial vehicle telematics technologies focus on reducing hard braking -- but hard braking can be very safe behavior if you're braking for a pedestrian or a car pulling out, and the biggest reason for hard braking is actually driver distraction. We detect the distraction which takes place seconds before the hard brake and the dangerous situation. 

Nauto can also detect imminent collisions when a driver is about to rear-end a vehicle or strike a pedestrian or bicyclist. Accuracy and operating at night are critical for these risks since many collisions take place in crowded urban environments and during lower visibility. Many other technology providers claim "forward collision warning" but don't operate at night, or only within a narrow speed band (not too slow and not at highway speed), or can't detect pedestrians. This is where the quality of the AI matters much more than many of the "check the box" knockoff cameras we see on the market that claim they can do something some of the time under the perfect conditions. Collisions don't happen in perfect conditions. You need AI and Nauto precisely when things get dicey. 

GN: How does the U.S. differ from Europe in its approach to transportation and regulation, and what are some areas where the country is falling behind or could improve?

Stefan Heck: Europe overall has a collision rate and especially a fatality rate that is about 1/3 of the US (2-5 fatalities per 100,000 populations vs 12.4/100K population annually). So why is Europe better? There are many contributing factors: EU cars are designed to protect pedestrians more and do less damage since they are smaller and lighter on average. Many EU cities provide more and separate space for pedestrians and bicyclists, especially in urban areas. Insurance costs are also substantially lower since many medical costs are covered by national health plans rather than motor insurance. Europe also adopts new safety standards faster -- e.g. the EuroNCAP and safety regulations that require ALL new cars sold in 2024 onward to have pedestrian collision warning, distraction and drowsiness warnings -- essential capabilities like Nauto's system in ALL new vehicles. Distraction is estimated to kill around 1500 people per year in the US, but our own data shows it's probably closer to half of the 40 000 fatalities a year in the US. Europe is tackling this problem directly. By contrast, the US has focused on backup cameras, which saved 58 - 68 lives per year, worth doing, but not our biggest safety problem.

GN: Can you give us your predictions for what fleets will look like five to ten years from now? What technologies will be standard, how will the regulatory regime change, and will humans still be behind the wheel in most cases? 

Stefan Heck: In 5 years, we won't see massive changes. AVs will still be nascent - operating mainly in highly mapped, sunny, constrained routes or at low speeds. The installed base of vehicles on the road only turns over every 12 - 15 years. Hence, while many ADAS (collision warning) and DMS (distraction and drowsiness alert) systems will become standard on new vehicles, the overall adoption grows gradually. This is likely to be very different in smart, safety-conscious fleets who are already embarking on complete retrofits of AI-powered safety technology today and pulling these savings and safety benefits forward by 5 - 10 years. 

In 10 years, we will begin to see significant adoption of both EVs (electric vehicles) and AVs (semi-autonomous or what is known as level 2/3 or fully autonomous known as level 4/5) will become much more common. But AV adoption will not be evenly spread. First, they will remain expensive, so they remain a luxury item. Some segments like long haul trucking on interstate highways lend themselves much more to autonomy, even for commercial fleets. But even they may not be 100% driverless -- just like airlines today fly 95% of the flight on autopilot, but the pilots often still land and take off by hand. 

Many people both on Wall Street and in the Tech industry miss that many commercial vehicle segments are not likely ever to go AV because the driver is an essential part of the service provided. 80% of Nauto's 700 fleets are in segments where the person in the vehicle is a core part of the service when the vehicle gets to the customer, for example, a utility line person, an HVAC service technician, or pest control expert, or an electrician. Even for many delivery fleets, the couriers are also the sales force, customer service, and a crucial part of the customer experience. Imagine UPS without "brown" -- the friendly person who smiles, replaced by an insect noisy drone dropping off the package on your doorstep. Amazon has backed away from many of its drone delivery claims as it realized that humans do many more things for delivery than just carry a package. If the technician or courier is going to be in the vehicle, the fleet will be paying their salary. Hence, an AV's labor and time savings over an AI safety solution available today are negligible.