SAE “Level 2+”: Where Future Driverless Ambitions Meet Today’s Market Realities

At CES 2019, the automotive industry continued to dominate, with all of the core automotive technology megatrends—connectivity, automation, electrification, and shared mobility—on display. However, while the automotive industry’s focus on the world’s largest consumer technology event shows no sign of slowing, there was a marked change in the overall focus, with the driverless prototypes that defined automotive at CES 2018 largely displaced by demonstrations of next-generation infotainment technologies and the repeated refrain of “Level 2+.” Having demonstrated their commitment to and investment in the smart mobility future, Original Equipment Manufacturers (OEMs) are now keen to emphasize how they can begin to monetize these investments in the difficult market conditions of the present.

Between the smart mobility announcements of CES 2018 and the more short-term focus of CES 2019, driverless vehicles have come to market commercially for the first time, in the form of Waymo One, a small-scale commercial operation, offering trips to 400 early adopters in a geofenced area of Phoenix, Arizona. To OEMs, the modest beginnings of Waymo’s commercial offering brought into sharp relief some important truths.

First of all, the driverless mobility revolution will begin not with a bang, but with a whimper. Scaling these services geographically will take time, with widespread availability years out, and profitability an even more distant prospect. This slow start has even applied to Waymo, the primary competitor behind which OEMs continue to trail in driverless vehicle development. While the case for smart mobility to deliver considerable improvements in safety and efficiency is by no means diminished, considerable shifts in consumer trust and legislation will need to be achieved before smart mobility services can begin to scale.

In the meantime, OEMs and their suppliers must begin to justify the billions that they have invested in driverless vehicle technologies in their race to beat Waymo to market. While OEMs will go to the ends of the earth to shave a few cents off raw material costs, they have thrown considerable funding behind investments in Artificial Intelligence (AI), new sensors, Vehicle-to-Everything (V2X), high compute platforms, and high-bandwidth Intelligent Vehicle Network (IVN) technologies. However, in the context of a slowing passenger vehicle market, the signs of belt tightening are clearly visible, with Magna leading the calls for restraint, and rumors abounding of a consolidation of resources in the German automotive market.

On top of the setbacks for Level 5 automation, the launch of the AI Pilot system on the Audi A8 has demonstrated that many markets are not even ready for Level 3 functions. The obvious response has been to take hardware and software components developed for fully driverless vehicles and target them at “Level 2+,” enabling a set of functionalities that fall within the Society of Automotive Engineers’ (SAE) Level 2 boundary, but in a way that is more robust than contemporary offerings. Furthermore, by “overachieving” with respect to hardware in Level 2+ implementations, there is considerable scope for Over-the-Air (OTA) updates to deliver a more comprehensive autonomous experience as improvements to software and legislation allow.

As OEMs and suppliers look to rationalize their driverless platforms for short-term monetization, it is interesting to note which of the many technologies (hardware and software) that have been positioned to enable driverless operation have been deemed necessary to give today’s semi-autonomous systems that extra “+” and which are clearly considered as more of a long-term opportunity.

Examining the systems demonstrated by NVIDIA and Mobileye, it is quite clear that vision-first approaches are preferred, with varied supporting sensors, AI, 360-degree perception, computing power far beyond today’s Level 2 implementations, centralized computing architectures (in opposition to smart sensors), and Highly Automated Driving (HAD)-grade digital maps have all made the cut. Conspicuously absent are 5G, V2X (802.11p or 3GPP Rel. 14), and mechanical Light Detection and Ranging (LiDAR).

The gulf that exists in hardware and software complexity between contemporary Level 2 deployments and the Level 2+ systems is significant, and serves to illustrate the vast spectrum of functionality that can be neatly filed away under the definition of SAE Level 2. It also goes some way toward explaining why contemporary Level 2 systems have proven so fallible, becoming a source of frustration for many consumers, and enabling a low level of autonomy in a way that is safe, consistent, and robust, which has proven to be a much more complex task than previously considered.

Clearly, technology component suppliers (both hardware and software) who have established their value proposition in the context of Level 5 driverless robotaxi applications must recognize the need to radically reposition their offering to gain traction in the Level 2+ market. The product-selling economics of the traditional passenger car market are far removed from the recurring service revenue streams of the driverless robotaxi market. Hardware component suppliers, including sensor and compute platform developers, must be conscious of OEMs’ concerns about cost and energy efficiency in the passenger vehicle market, while software module providers and AI specialists must renew their focus on compliance with functional safety requirements.