NHTSA’s release last month of crash data regarding automated (ADAS) and autonomous (ADS) technology was not revelatory – it neither condemns the emerging technology nor confirms it as ready for widespread adoption in the consumer vehicle market. Instead, it serves as a reminder that transformative technologies that seem obvious in hindsight are not so obvious when first introduced. What is commonplace today was at one time new, novel, and maybe even a little unnerving.
Today the airbag, Electronic Stability Control (ESC), and Automatic Emergency Braking (AEB) are standard features on nearly every car produced and sold in the United States. Over time, data has proved their net benefit to overall safety. But their ultimate success was not a foregone conclusion. These products took a long and winding path from inception to acceptance, some over many decades. Reviewing the history of these technologies and their varied path to commercial acceptance will help shed light on the path forward for autonomous driving technology.
While airbags are ubiquitous today – America’s best-selling sedan, the Toyota Camry, now comes standard with 10 airbags – their path to widespread adoption was fraught, nearly failing multiple times along the way. The progenitor of the modern airbag was first patented in the 1950s, yet it took another 40 years before airbags achieved widespread acceptance as a necessary safety device.
The earliest challenges were technological. Initial automotive airbags were unable to inflate the bag quickly and forcefully enough to provide a sufficient crash cushion. Breakthroughs were made in the 1960s with the invention of crash sensors and the use of propellant to inflate the bags.
The next challenge was consumer adoption. General Motors became the first automaker to make airbags available in production cars with 1974 model year vehicles. However, they did not sell – only 10,000 cars equipped with airbags were purchased in three years before GM stopped offering the feature. Airbags were not made available again until Mercedes-Benz reintroduced them in 1984, along with regulatory developments encouraging airbags. In 1991, Congress passed a law mandating airbags on both the driver and passenger sides for all new cars and light trucks effective in the 1998 model year.
Despite being a safety innovation, the airbag technology itself raised new safety concerns. In order to protect normal-sized adult occupants, airbags deploy with a lot of force, which can be harmful to children and even smaller adults, particularly when seated in unusual positions. Over time, new technological advancements, educational campaigns, and legislation addressed these risks, seriously mitigating the negative externalities of airbags.
Ultimately, the widespread adoption of airbags required an extraordinary combination of technological innovation, changing consumer sentiment, and evolving regulatory actions and legislative mandates over the course of four decades.
Electronic Stability Control
Electronic Stability Control (ESC) uses computer-controlled braking of individual wheels to improve vehicle stability during extreme maneuvers, helping keep vehicles on the roadway to avoid single-vehicle crashes and rollovers. In contrast with airbags, ESC achieved widespread adoption quickly as it offered exceptional safety benefit without introducing new safety risk.
First introduced in the mid-1990s, ESC drew attention after a Mercedes A-Class flipped onto its side during a “moose” test, which requires a vehicle to swerve around a fake animal without flipping or rolling over. Mercedes subsequently made its Electronic Stability Program a standard feature in all its models.
The 1990s saw the advent in the United States of Sport Utility Vehicles. The higher center of gravity of SUVs made them less stable in close turns and more susceptible to rollovers. The failed “moose” test, followed by the Firestone defect in Ford Explorers that led to hundreds of deaths, highlighted these risks.
The industry responded, as car companies increasingly installed ESC into vehicles, NHTSA added a dynamic rollover test to its five-star safety ratings program, and the safety community began to study the effectiveness of ESC in keeping vehicles on the road and preventing these fatal crashes. By 2006, the value of ESC was incontrovertible, and NHTSA proposed to mandate ESC in all new light duty vehicles. By that time, according to NHTSA’s proposal, close to a third of model year 2006 light vehicles had already been equipped with ESC. The requirements were finalized in 2007, ensuring that by model year 2012 all new vehicles would be equipped with ESC. NHTSA reported that “ESC has among the highest life-saving potential of any vehicle safety device developed in the past three decades, ranking with seatbelts and air bags.”
While the 17-year time span between the initial introduction of ESC systems and full adoption may seem glacial in the tech environment, this was lightspeed for an automotive safety advancement. The US motor vehicle fleet includes about 275 million vehicles, making it difficult for a new technology to show statistical effectiveness until it has already proliferated through the fleet. Model vintages commonly last for many years without a refresh or replacement. In addition, the need to demonstrate effectiveness in addressing a known safety risk means that regulatory mandates often trail the deployment of new technology. ESC’s rapid proliferation through the fleet was aided by both the early evidence of its extreme effectiveness and the lack of negative externalities associated with it.
Automatic Emergency Braking
Automatic Emergency Braking (AEB) and its sibling technology, Forward Collision Warning (FCW), have been estimated to reduce incidents of rear-impact crashes by as much as 50% by detecting when a possible collision is about to occur and stepping in if the driver fails to brake effectively to avert the crash.
They first appeared on luxury cars in the mid-2000s and have been broadly adopted since despite the known shortcomings of AEB technology. While useful, known performance limitations include inclement weather and false positives that can startle drivers by triggering sudden and necessary activations. The adoption push reflects the willingness of NHTSA and the automotive safety community to weigh the safety risk of new technology against its benefits, and to tolerate infrequent and relatively minor risk when warranted. In fact, the consensus view on the risk associated with AEB is that “these occurrences are expected to be infrequent and have a minimal influence on the greater benefits of these systems."
NHTSA also initially took the position that there was a faster and better path to promote the development and adoption of transformative technology than the regulatory path. In 2016, NHTSA and the Insurance Institute of Highway Safety secured voluntary commitments from all the major automakers to install AEB in virtually all their new vehicles by September 2022 – an estimated three years earlier than if it had pursued regulatory mandates, preventing 28,000 crashes and 12,000 injuries during that time frame. By the end of 2019, four manufacturers had already fulfilled their commitment and many others were about 75% of the way there.
NHTSA later justified the approach by pointing out that “because many modern vehicle safety technologies are software-controlled and still relatively new, they are evolving very quickly. Standard setting at this early stage of technological evolution must be undertaken with great care, given the risk of inadvertently stymieing innovation and stalling the development and introduction of successively better versions of these technologies." Even without a regulatory mandate, NHTSA has been using its authority to investigate potential safety defects and to force recalls when failures occur more frequently or with more serious safety consequences than expected. Last year, the Bipartisan Infrastructure Bill directed NHTSA to mandate AEB, and NHTSA has made clear it will do so shortly.
The NHTSA data release demonstrates that current ADAS and ADS are not perfect, but the systems will continue to evolve and improve. Breakthroughs in artificial intelligence, such as that driving the Ghost Autonomy Engine, will allow these technologies to scale through the new vehicle fleet and eventually to be ubiquitous. While there may be some challenges to overcome in the meanwhile, over time AV technology can secure its place in history as perhaps the most impactful automotive safety advancement ever.
AV technology is following a well-worn path of previous transformative safety features, most notably those described above. What implications does the history of these previous technologies portend for the future of autonomous driving?
First, achieving safety benefits often takes multiple breakthroughs and ongoing investment. The progenitor of modern airbags could not deploy with enough speed and force. Today’s ADAS require ongoing driver monitoring and/or attention. ADS requires next-generation AI to make it scalable to a broad consumer vehicle market. As with AEB, software-based autonomy can iterate and improve rapidly, and regulatory mandates bear the risk of locking in technology before the safety benefits are fully achieved.
Second, the fewer the externalities, the faster autonomy will be adopted. ESC did not introduce new safety risks, unlike the airbag technology that preceded it. ESC began to deploy quickly, and its effectiveness at averting single-vehicle crashes and rollovers allowed it to become a required safety feature without controversy. The good news from the available data is that ADS technology does not appear to be introducing new or unusual behaviors that cause serious accidents on our roads, minimizing the risks of broader adoption.
And finally, the perfect should not be the enemy of the good. Despite its shortcomings, the cost-benefit analysis for AEB was clear – the upside greatly outweighs the downside. The same is true for developing ADS technology. Autonomy doesn’t have to be perfect everywhere for society to begin benefitting from it. Even limited applications, such as highway-only Level 4 autonomous driving, can dramatically enhance the driving experience for many and contribute to improving overall road safety. As these systems mature, they can expand over time to more driving domains and use cases, ushering in a new era of safer personal transportation.