The automotive industry, long defined by combustion engines and mechanical prowess, is undergoing a seismic shift.  Arm is playing a central role in enabling the automotive industry’s fundamental shift towards software-defined and AI-driven vehicles through several key strategies and technologies.

In a recent interview with EE Times, Dipti Vachani, senior vice president and general manager of the Automotive line of business at Arm, offered a compelling perspective on this transformation, highlighting the pivotal role of software and the collaborative efforts driving this evolution. 

According to Vachani, the vehicle is rapidly transitioning from a “hunk of metal” to a “software-defined, ever-evolving dynamic machine.” This profound change, fueled by electrification, advanced driver-assistance systems (ADAS), and burgeoning in-vehicle entertainment (IVI) demands, presents immense opportunities and significant challenges for the industry.

Tectonic shifts in automotive evolution

Vachani underscores that the automotive landscape is experiencing its most significant upheaval in decades. Three primary forces are propelling this change: 

First, the increasing adoption of electrification is altering powertrains and necessitating sophisticated software for battery management and overall vehicle efficiency. 

Second, the proliferation of ADAS equips vehicles to “think” and enhance safety through features like warnings and automated assistance. Vachani clarifies that this represents a gradual progression towards safer vehicles, not necessarily an immediate leap to full autonomy. 

Finally, rising consumer expectations for in-vehicle entertainment (IVI) are transforming the car cabin into a digital experience space, particularly as vehicles become safer and potentially more autonomous. 

The emergence of new car manufacturers globally, particularly from the U.S. and China, further accelerates innovation.

Arm’s strategy in a transforming industry

Arm finds itself strategically positioned in this dynamic environment to capitalize on these transformative trends. Vachani emphasizes several key assets that provide Arm with a unique value proposition. 

Low-power computing is paramount, not just for electric vehicles to maximize range but also for traditional cars to improve fuel efficiency. Vachani notes, “Despite this increase in desire for compute, we still have to do it in the lowest power possible. So our low-power compute is so important to this transition.”

Arm also offers heterogeneous compute capabilities, encompassing CPUs, GPUs, AI accelerators, and image signal processors (ISPs). 

This diverse portfolio is essential for handling modern vehicles’ increasingly complex computational demands, including graphics, artificial intelligence workloads, and processing data from numerous sensors. 

Another crucial advantage is cloud-to-car parity, which utilizes the same underlying architecture in cloud computing environments and within the vehicle. This consistency enables original equipment manufacturers (OEMs) to deploy and update software reliably in real-time. Vachani highlights Arm’s “tremendous journey in cloud computing and increasing our footprint” as a key enabler for this parity.

Recognizing the critical importance of safety in the automotive sector, Arm transitioned to an “automotive enhanced portfolio” approximately five years ago, providing Automotive Safety Integrity Level (ASIL)-certified intellectual property (IP) for safety-critical applications. 

This provides a robust foundation that customers can rely on, eliminating the need to “reverse engineer and create safety.” 

These strategic assets have propelled Arm to a significant market penetration, with 94% of global OEMs utilizing Arm technology in their vehicles for ADAS or IVI functionalities. As Vachani states, “those three assets that Arm has enabled us to grow to a point where 94% of global OEMs are taking advantage of Arm in their car in some way.”

Complexity of ADAS and the autonomous future

The evolution of ADAS is a key stepping stone towards fully autonomous driving. While the technology for autonomous vehicles is advancing rapidly, 

Vachani points out that the major hurdle lies in human acceptance and the societal expectation of near-perfect safety. She regretfully remarks, “We will not forgive a robot for making a mistake; we will forgive a human for making a mistake.” 

Interestingly, the emergence of generative AI is a valuable tool in expediting the development and validation of autonomous driving systems. 

Vachani explains, “We can now create scenarios with generative AI, right? We’re not necessarily limited by the vehicles’ driving streets and their hours on the street. Generative AI allows us to create and simulate that vehicle’s scenarios.”

Vachani anticipates that initial deployments of fully autonomous technology will likely be in fixed routes, such as small shuttles operating in controlled environments with limited variables. 

Universal autonomous driving faces a more significant challenge due to the sheer complexity of variables and the stringent zero-tolerance for errors. Despite statistical projections suggesting that autonomous driving could eventually be statistically safer than human driving, public perception and acceptance of robot errors remain critical challenges.

Level 1 and low-end Level 2 systems dominate the current landscape of vehicle autonomy. However, Vachani predicts that L2+ vehicles offer near-Level 3 capabilities where a human can intervene.

Still, the vehicle can handle most driving situations on highways and city streets and will become the highest-volume vehicle in the next four to five years. 

Government regulations distinguish certified Level 3 autonomy from higher levels of autonomy (L4/L5), which will see more restricted initial deployments, potentially within geofenced areas.

Collaborative path for software-defined vehicles

Recognizing the significant software challenges inherent in the automotive transformation, Arm spearheaded the Scalable Open Architecture for Embedded Edge (SOAFEE) initiative approximately four years ago. 

This collaborative effort directly responded to feedback from OEMs who identified key barriers to deploying software-defined vehicles at scale. These critical requirements were the ability to use and port the same software across different hardware platforms in cars ranging from low to high cost, ensuring software consistency between the cloud and the edge (in the car), and the capacity to develop software before hardware availability to keep pace with the growing demand for AI-powered capabilities. 

As Vachani succinctly puts it, SOAFEE aims to solve the challenges of “Being able to use and port the same software across different hardware platforms in all cars from low to high cost; Having software consistency in the cloud and at the edge (in the car); and Keeping pace with the growing demand for AI-powered capabilities throughout the vehicle by being able to develop software before the hardware becomes available.”

SOAFEE has evolved into a robust consortium, bringing together over 140 members. Recent additions to the SOAFEE community include prominent OEMs such as Geely, General Motors, and Tata Motors, expanding the collaboration to over 120 leading industry players. 

Vachani emphasizes the “unprecedented collaboration” that has yielded significant progress over the past three years. This collective effort has already fostered a “brand-new ecosystem of software solutions that will be critical to enabling Arm Compute Subsystems (CSS) for Automotive in 2025 through delivering software consistency to support the silicon development and deployment process”.

Beyond SOAFEE, Vachani highlights the SDV Alliance, of which SOAFEE is a founding member, as another crucial opportunity for broader industry collaboration to advance the software-defined future of automotive.

Regional Variations and Strategic Partnerships

While the fundamental need for low-power, heterogeneous computing in central computing and ADAS/IVI systems remains ubiquitous across different regions, Vachani notes some regional variations, particularly in powertrain technology. 

Regulations also play a significant role in the deployment of autonomous driving. Europe currently faces more complex regulatory hurdles compared to the U.S. and China, where trials and even some commercial deployments like robotaxis are underway in specific cities. 

Strategic partnerships are also instrumental in driving progress. Arm’s collaboration with Nvidia is particularly significant, with Nvidia’s autonomous, robotic, and automotive solutions, including Thor and Orin, being built on Arm technology. 

As Vachani remarked after the recent GTC conference, “GTC was a great time for us because we saw our partnership at Nvidia come to life. All of their autonomous solutions, as well as their robotic solutions and their automotive solutions, that is, Thor and Orin, are built on ARM technology.” 

Architecting the automotive future

Dipti Vachani’s insights underscore the profound transformation sweeping the automotive industry, with software taking center stage. Arm has established itself as a central enabler of this revolution with its strategic focus on low-power and heterogeneous computing, cloud-to-car parity, and safety-certified IP. 

The SOAFEE initiative, driven by unprecedented industry collaboration, addresses the complexities of developing software-defined, AI-enabled vehicles.

While the path to fully autonomous driving presents significant societal and technological challenges, advancements like generative AI and a phased deployment approach offer a promising way forward. 

From EETimes