Autonomous drive by 2020: Over-promised, under-delivered?
A look at the future vehicle through the eyes of SAE International
This article is the first in a series of six articles that provide a practical look at the feasibility of connected vehicles and autonomous driving.
2020. Four short years from now. So close that we can imagine being there, yet distant enough for major changes to occur. In some circles, 2020 is the year there will be an unprecedented automotive advancement – the debut of self-driving vehicles that flawlessly chauffer us around. We’ve heard this narrative over and over in countless news articles, conference speeches and public announcements. However, is it actually feasible? The answer is “Maybe” – but likely not quite how it is predicted.
When you think about the autonomous vehicle ecosystem – consisting of automakers, technology providers, lawmakers, standards providers, infrastructure developers, and, last but not least, drivers – there are hundreds, possibly thousands of hurdles to overcome before these vehicles can hit the road in any form. Removing control of the vehicle from a human driver turns our current system of transport on its head. The industry is currently grappling with the task of re-imagining everything from roadways to vehicle interiors in an effort to develop an experience that is safer than traditional driving and is viable with technologies currently available. Easier said than done, particularly with only four years to go.
What it’ll take
So, how exactly can the automobile of the future live up to the hype, whether in 2020 or beyond? According to SAE International, the answer is through carefully organized collaboration. Instead of a single contributor, such as an OEM, weighing in on areas outside of their expertise, the logical answer is to hear input from many voices and collectively adopt a path forward that can be replicated across the globe.
Jack Pokrzywa, SAE’s Director of Global Ground Vehicle Standards, explains, “If you think in musical terms, SAE provides the stage to the musicians (standards developers) of the orchestra (committee), supplies sheet music (rules and policies), and ensures that the effort is conducted and monitored by the industry, government, and academia experts. The technology to enable driverless technology reaches far beyond traditional automotive, requiring input and acceptance from numerous sources. It will change the way our world operates, so we need to get it right.”
The idea of “getting it right” is extremely complex – far beyond taking vehicle control away from the occupants. The concept of driverless vehicles dates back to the 1920s but only recently has become a potentially feasible undertaking.
The introduction and steady integration over the last few decades of advanced driver assist systems (ADAS), like parking assist and adaptive cruise control, along with improvements in GPS technology and mobile communications, have laid the initial groundwork for fully autonomous driving. Coupled with the post-recession auto industry boom and significant advances in software and hardware development, most automakers began developing self-driving technologies.
However, a key requirement of autonomous driving is for a “driverless” vehicle to be safer than its human driver. This task is proving to be very difficult to achieve, as there are aspects of the task that extend well beyond manufacturing and technological capabilities. Not only must a vehicle be able to gather accurate and detailed information on the world around it, it must also make independent, educated, and quasi-moral decisions based on that information.
As of now, the industry is still working on a solution that checks each of these requirements and finding that the task requires a significant degree of collaboration to ensure personal safety and data protection, not to mention operational efficiency.
Taking on Silicon Valley
Seeking answers (and, to be honest, anxiously attempting not to be overtaken in the race to autonomous vehicle production), automakers are looking to Silicon Valley, the epicenter of disruptive innovation, as a roadmap for future success. To secure their place in this forward-thinking culture, automakers are restructuring their operations and processes, sometimes even acquiring or partnering with the companies they strive to emulate.
What original equipment manufacturers (OEMs) are overlooking is that Silicon Valley might know how to develop technology solutions quickly, but sometimes at the expense of reliability. Think of it this way: every year new smartphones with advanced features are released after much hype and speculation. To deliver on their promises, the highly awaited phones often debut with their expected features. The tradeoff is the occasional bug in the hardware or software that is corrected later with an update.
This is how Silicon Valley can innovate so quickly – with iterative design, it relies on the user to identify potential issues and fixes them in subsequent software and hardware releases. When scaled to a full-size automobile responsible for the safety of its occupants, this becomes a worrisome proposition. Iterative design is no longer an option. Both the software and hardware have to work flawlessly upon launch — there is no room for “404 Error” when it comes to human safety.
What if the hype is just … hype?
Unlike a favorite smartphone app, imperfect vehicle technologies can introduce risk – the most severe being collisions with other vehicles, buildings, and even pedestrians.
Despite these risks, automakers are still racing to bring various degrees of autonomous driving to market first. We’re already seeing the beginnings of this trend with Tesla’s AutoPilot, Nissan’s Piloted Drive, and Cadillac’s upcoming Super Drive.
SAE categorizes these technologies as Level 3 “Conditional Automation” automated driving systems. Level 3 falls in the middle of SAE’s Levels of Driving Automation for On-Road Vehicles. With the goal of establishing common terminology among the differences of automated driving, SAE’s J3016 standard rates vehicle technologies on a scale of 0-5 ranging from “no automation” to “full automation.” Each category is defined by the responsibility of driving tasks as assigned to a human or the system in certain driving modes.
The difficulty of a Level 3 vehicle is the heavy reliance on both humans and the onboard technology. In the past months, we’ve seen the first instances of documented crashes from the use of semi-autonomous technologies. Most of these crashes have been a result of humans not reacting appropriately when the task of driving is transitioned from the system to the driver. This transition occurs when driving conditions suddenly change or the system encounters a malfunction.
The information SAE has collected suggests that the handoff of responsibilities occurring in Levels 2 and 3 will be one of the largest difficulties to overcome as 2020 nears, so much so that many industry experts question if true Level 3 vehicles are possible, let alone feasible. This is an issue beyond the capabilities of Silicon Valley technology development, as the technology cannot control human reaction.
In the autonomous technologies already on the road, we’ve already witnessed vehicle hacking, features that randomly auto-disable, and cases of false positives and false negatives resulting in erratic driving. Some have resulted in injury and death. OEMs are grappling with how to solve these issues before the consumer develops a negative connotation to the technology, making the hopes of greater advancements in autonomous driving dead on arrival.
Then, there’s the issue of introducing autonomous technologies that are even more advanced than Level 3. Companies such as Uber and Ford have announced their intentions to have fully autonomous vehicles requiring minimal human interaction. While SAE expects these technologies to be safer in the long run than those that share driving responsibilities with a human, they require greater sensor technology, processing power, and supporting infrastructure.
We’re just scratching the surface on developing a communication system that connects a car to its surroundings. The same goes for sensors and processors that can accurately identify hazards in and around the road. And issues related to data storage, analysis, and access? Few, if any, have considered the ramifications of the introduction of cars that are essentially supercomputers in the collection and use of personal data.
2020 is beginning to look a lot different that it’s been made out to be.
A common vision
When asked about the significance of introducing autonomous vehicles, SAE’s Pokrzywa noted, “This is perhaps the biggest change in the transportation industry since the automobile co-existed with the horse and buggy. It will require a cultural change and insight from many different parties, not just advanced technology. The new solutions will have to be developed collaboratively in a global context unlike ever before.”
With so many unknowns yet to be addressed, the 2020 goal set by so many becomes highly unrealistic. Historically it takes years – sometimes decades – to successfully bring a single technology to market. Just look at the timeframe to bring airbags and antilock braking systems into widespread use – and neither of those systems required the infrastructure changes on which fully autonomous vehicles will rely.
To speed this process, SAE is calling for a global web of stakeholders to be established between OEMs, suppliers, government agencies, technology providers, and infrastructure engineers to create a well-vetted plan for moving forward.
As the single unbiased, global organization capable of bringing together each global stakeholder, SAE is currently leading the development of standards that address the issues standing in the way of successful autonomous vehicle deployment. In fact, the organization has been working behind the scenes for many years, laying the groundwork for upcoming driverless technologies.
SAE is looked to as a creator of universal descriptions for core driverless technologies, with more than 55 industry coalitions to identify and overcome roadblocks that stand between vehicles of today and fully automated driving. Just recently SAE’s Levels of Automation, J3016, were adopted by the National Highway Traffic Safety Administration (NHTSA) to guide the future development and categorization of autonomous vehicles. SAE is also coordinating with global powerhouses like the International Organization for Standardization to expand its levels of automated driving and other expert insights into autonomous reliability and safety across the globe.
So, what’s next?
The key to success is balancing hype with reliability. If automakers rush development and continue racing each other to their 2020 goal, the technology could literally crash and burn before it ever sees Level 5 – full automation. At SAE, as advocates for the advancement of mobility through engineering, that is the last thing we’d want to see.
Alternatively, we’re calling for collaboration between stakeholders to intelligently approach the plethora of issues that need to be solved. There is still much to be done in developing roadways, communication systems, laws, technology standards, and educating consumers before these technologies become viable.
SAE collectively believes that 2040 may be a more reasonable goal to deploy these technologies than 2020. It is worth noting this could happen sooner, and it could happen later. It all depends on how fast the industry adapts and adopts a new way of considering transportation.
There is a need for necessary cultural change, stakeholder collaboration and advancement of core ADAS technologies that four years cannot provide us, and while it may not be what automakers and consumers want to hear, it is a reality as engineers develop these systems.
SAE’s practical approach may not be the most popular, but it is based on decades of research, input from thousands of key stakeholders, and systematic evaluation. The organization has seen its fair share of hype over the years. Hype comes and goes, yet the need for practical and safe mobility for our world remains.
Society of Automotive Engineers (SAE) International
1. SAE J3016 – “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles” can be downloaded free of charge at.