Safety first

For over a century, the success and safety of automobiles have depended on a highly sophisticated sensing and computing solution, one that fits comfortably behind the steering wheel: their drivers. The driver, and the driver alone, was called on to stick to the speed limit, abide by the traffic rules, engage in difficult and potentially consequential maneuvers, chart the optimal course to the destination, and safely pilot the vehicle along its journey.

The cars we drive today are meticulously designed around our innate human skills and aptitudes, which we continually perfect throughout our driving lives. But with the World Health Organization reporting 1.3 million road fatalities each year and near misses more common still, humans can hardly claim to be infallible behind the wheel. Easily distracted, prone to stress and fatigue, victims of our own egos, and worse at reading maps than we would like to admit, there continues to be plenty of room for improvement.

Many past enhancements in traffic safety have been the result of mechanical changes or additions to the car: seatbelts, baby seats, airbags, optimized designs informed by crash testing. The current wave of improvements goes beyond that. As a result of the massive progress in sensing and computing hardware, wireless communication, data processing, and algorithmic sophistication, technology has matured to the point that it can begin to overcome some of our human limitations.

The bottom line? In addition to saving hundreds of thousands of lives, the financial benefits of automation add up. Back in 2018, Swiss Re and Here estimated that widespread adoption of advanced driver assistance systems could cut global car insurance premiums by US$ 20 billion by 2020. And with the PwC strategy & consulting team estimating the global value of the automated driving technology market to hit US$ 270 billion by 2030, automakers and their suppliers around the world are competing for their slice of that pie.

Stay informed

ADAS explained

But first, what are advanced driver assistance systems (ADAS), what functionalities do they offer, and from what point on, precisely, do they classify as advanced? According to the Society of Automotive Engineers (SAE), cruise control, which lets a vehicle maintain a set speed, falls short of being considered “advanced” (see infographic). Because it only assists drivers by controlling a single functionality – here, speed – it’s merely a “driver assistance system.” Driver assistance systems that control both speed and steering, on the other hand, are considered advanced.

The list of technologies that make up ADAS solutions is long – longer still if you account for the diversity of marketing terms automakers use to describe their proprietary implementations of roughly the same feature set. To simplify the ADAS technology landscape, the American Automotive Association has proposed a terminology comprising twenty technologies, which can be grouped into five categories:

1. Automated driving tasks (Adaptive Cruise Control, Dynamic Driving Assistance, Lane Keeping Assistance)
2. Collision alerts (Blind Spot Warning, Forward Collision Warning, Lane Departure Warning, Parking Obstruction Warning, Pedestrian Detection, Rear Cross Traffic Warning)
3. Collision mitigation (Automatic Emergency Steering, Forward Automatic Emergency Braking, Reverse Automatic Emergency Braking)
4. Parking assistance (Fully-automated Parking Assistance, Remote Parking, Semi-automated Parking Assistance, Surround View Camera, Trailer Assistance)
5. Miscellaneous driving aids (Automatic High Beams, Driver Monitoring, Night Vision)

Each of these technologies represents a stepping-stone on the path towards higher degrees of assistance, and then towards full autonomy. The SAE represents this as a progression from vehicles that only provide warnings or momentary assistance (Level 0) to vehicles that drive autonomously in all conditions (Level 5).

^{© SAE}

A safer, more convenient ride

For all the hype around automated driving, the road leading us there will likely be long and bumpy. No doubt, progress has already been impressive, with each new feature – blind spot warnings, automatic emergency braking, etc. – doing its part to increase road safety for those inside and outside the vehicle. Still, with a few lonely exceptions, even the most advanced cars for sale today offer Level 2 automation at best. While a growing number of vehicles may have the technological chops needed to shuttle passengers around, regulators continue to be wary of letting drivers take their attention off the road environment – the hallmark of Levels 3 and above.

In other words, in the SAE’s chart, we are still in “blue” territory: Levels 0 to 2. Even in today’s most advanced Level 2 autopilots, sometimes referred to as Level 2.9, the driver is required to remain engaged and prepared to take control of the vehicle at all times, and, critically, assume responsibility in the case of an accident.

According to projections by ABI Research, Level 2 will continue to dominate passenger vehicles well into the future, with Level 3 and higher expected to make up just under 15 percent of new vehicle sales in 2030. By that time, roughly half of new vehicles will still classify as Level 1 at best. And mass adoption of the highly autonomous vehicles is unlikely to begin with consumers. Because of their higher cost of acquisition and repair, and because they will only be permitted in a restricted set of conditions, vehicles sporting automated driving features will likely enter the market via businesses and business-driven use cases.

The benefits to business

The business rationale behind highly automated vehicles is clear: reduced operational costs, increased safety, and optimized efficiency. At least in the early days, fleet owners will see the strongest case to invest in more costly highly autonomous vehicles. They will be in the best position to offset the otherwise high per-mile cost by maximizing vehicle allocation and keeping them on the road virtually non-stop while saving costs in the form of insurance fees, salaries for highly qualified drivers, or by eliminating the driver altogether, sometimes replacing them with remote supervisors.

Robo-taxis are one popular use case that maximizes vehicle utilization – and, consequently, return on investment – that several OEMs including Ford in the US and Didi in China are investing in. While, initially, they will largely be made up of vehicles featuring Level 4 autonomy and requiring a (remote) driver, as the technology matures, they are expected to graduate to driverless Level 5 autonomy. In addition to offering passenger services, these vehicles could engulf other use cases as well, automating the delivery of groceries, meals, packages, and other goods.

The trucking industry also has a strong incentive to adopt higher levels of vehicle autonomy, with Daimler, Einride, Embark, and Volvo among the big names in the mix. While it will still take a few years for driverless trucks to transport goods from coast to coast, fleet owners can benefit from a variety of already-available forms of assistance that increase efficiencies or otherwise save costs. Parking assist and reverse assist, for example, take over the wheel to deal with maneuvers that otherwise require highly qualified, expensive drivers.

As a result of the massive progress in sensing and computation hardware, wireless communication, data processing, and algorithmic sophistication, technology has matured to the point that it can begin to overcome some of our human limitations.

Platooning, where several trucks automatically trail each other in a convoy, reduces drag, saving fuel and cutting down CO2 emissions. While one pilot study carried out in Germany by DB Schenker, MAN Truck & Bus, and Fresenius University of Applied Sciences measured fuel savings of three to four percent, the US National Renewable Energy Laboratories demonstrated fuel savings between ten and 17 percent, depending on the truck’s position in the convey. And in ports, where vehicles operate in a structured and obstacle free environment, driverless terminal trucks already relay containers from the shipping yard to the container ship.

First successes, more in the pipeline

Next time you’re driving, look around. At least some of the vehicles around you today are likely to boast Level 2 autonomy. Tesla, one company that sells consumer vehicles with Level 2 autonomy, has been racking up miles with their autopilot, which gives drivers a sampling of what full autonomy will feel like. But because it still requires them to remain fully aware behind the wheel, the level of autonomy it offers drivers falls short of Level 3. It isn’t alone. Daimler-Benz, General Motors, and BMW all have vehicles with advanced Level 2 autonomy in production. And with the European New Car Assessment Programme (Euro NCAP) including many Level 2 ADAS features in their star-based safety rating, it’s safe to assume that these features will slowly begin to appear in mid- to low-segment vehicles as well.

While breaking into Level 3 has proven to be a real challenge, there’s little doubt that it will be a game-changer: starting at Level 3, the driver will be able to let go of the wheel and do something else. Until SAE Level 5 is attained, such eyes-off the-road, hands-off-the-wheel service will only be available when specific conditions are met, or, in autonomous driving jargon, when the vehicle is within its operational design domain (ODD) for that specific autonomous functionality.

In SAE Level 3, it will be up to the vehicle to judge, for example, whether the operating conditions for autonomous driving are fulfilled based on its geographical location, traffic density, the time of day, and other conditions such as which lane it is driving in. As a result, Level 3 vehicles will need to be sufficiently “self-aware” to determine whether the sensor data they are fed can be trusted, and, if so, to request the driver to take over control.

The first company to hit the consumer market with a Level 3 production car is Honda, which started selling its Honda Sensing Elite ADAS system in a Japan-only version of the Honda Legend sedan in March of this year. Granted, only 100 vehicles will be sold at first, and the restrictions on the use of the technology will still be quite stringent. Drivers can only let go of the wheel and engage in other activities when they are stuck in slow-moving traffic jams. In other scenarios, the car offers additional driver assist functionality.

Drive across Phoenix, Arizona, and you might come across vehicles completely lacking a driver. There, Waymo, the Alphabet company operating the fleet of driverless taxis, is offering what might be considered as “supervised Level 5.” What passengers do not see is that the cars can fall back to human support if they encounter a situation that they are not qualified or able to handle. Examples might include an unexpected, blocked road or a traffic warden directing traffic using hand signs. And in China, Baidu, the tech giant, and AutoX, an Alibaba-backed company, have rolled out commercial service in Beijing and Shenzhen respectively.

Safety first. Full autonomy? Later!

Look at the big picture, and the signs are clear: while the initial hype and optimism surrounding driverless vehicles has given way to a more realistic assessment of the situation, momentum towards vehicle automation to increase road safety continues to grow. Driver assistance is already preventing injuries and saving lives on our roads today. And as vehicles begin to take over, first in narrowly defined scenarios, then in increasingly all-encompassing ones, they will likely prove themselves to sense more, react faster, and drive more safely than we do. And, who knows, a few decades from now, it might be considered reckless to even let a human sit behind the wheel...

Stay informed