The barriers to self-driving cars are significant, from costs needing to come down to regulations needing to be clarified around certain self-driving car features. Talk of millions of self-driving cars on the roads within the next few years seems remote, depending on the definition you use, but Lexus, BMW, Apple and Google are all developing, or rumoured to be, automated technology.
Fully-driverless cars are still some way off but partially automated technology has been with us for many years.
While there is significant investment being made in driverless technology, manufacturers need to tackle a range of ethical and technical issues – chief among them safety.
“Safety is the highest priority for car makers we talk with, for both the obvious technology factors associated with autonomous systems controlling all aspects of driving, but also to ensure that human passengers can trust their automated driver. If consumers don’t trust the autonomous systems in their cars are safe, then mass market acceptance of this technology will be slow to happen,” says Lakshmi Mandyam, VP Automotive, Embedded & Automotive Line of Business, Arm.
Mandycam makes the point that development costs are increasing exponentially as, “the software complexity and volume for autonomous systems is rising dramatically. To put this into some perspective, it’s predicted that a Level 5 vehicle will require a billion lines of code. Compare that to a Boeing 787 Dreamliner, which ‘only’ requires 14 million lines of code.”
When it comes to the safety of autonomous vehicles, however, most of the accidents involving them have been due to human error, so the bigger issue for the industry is what should autonomous vehicles be doing to reduce accidents?
“We are in constant discussion with car makers and our extended automotive ecosystem, which comprises of the top 15 automotive chip makers that license Arm’s IP, about progressing toward fully-autonomous driving,” says Mandyam. “While we certainly talk about how we can address their performance, power and security requirements, most of the discussions we have tend to focus on safety.”
According to Mandyam, autonomous driving is expected to eliminate human error.
“If consumers don’t trust the autonomous systems in their cars are safe, then mass market acceptance will be slow to happen.” Lakshmi Mandyam |
“Ninety four per cent of all accidents are a result of driver error and so we expect fully-autonomous driving to significantly reduce the number of accidents and fatalities.”
Autonomous vehicles will be dependent on sensors to detect what is happening around them and today engineers are defining the right mix of sensors that need to be implemented – but they need to also take into account the costs and computing power required, both are limiting factors.
The other key to vehicle safety will be how the software handles unexpected situations. All self-driving vehicles will have to make many hundreds of decisions every second in order to make adjustments necessary to keep the driver safe.
Vehicles equipped with high levels of autonomy are expected to require 100 times more compute performance by 2024 than is currently the case.
Car makers need to ensure that when it comes to the deployment of autonomous vehicles they are able to provide a safe and efficient compute platform.
“That is why safety cannot be an afterthought or be relegated when it comes to developing autonomous-class SoCs and systems,” says Mandyam.
“Unfortunately, the path to level 5 autonomy has tended to be paved with prototypes, often based on power-hungry, expensive data centre CPUs which lack even the most basic functional safety features.”
Prioritising safety
Arm has sought to prioritise safety over many years and that is why, according to Mandyam, the company’s IP is now in 65 per cent of the silicon used in ADAS applications.
“Our automotive ecosystem has access to the industry’s broadest array of functional safety IP with the latest ISO certifications,” she explains.
In fact, Arm’s Safety Ready programme encompasses not only existing safe but new and future products which have been through a rigorous functional safety process, including systematic flows and development in support of ISO 26262 and IEC 61508 standards.
Safety Ready is a one-stop shop for software, tools, components, certifications and standards which is intended to simplify and reduce the cost of integrating functional safety. By taking advantage of the programme’s offerings, partners and car makers are assured that their SoCs and systems will incorporate the very highest levels of functional safety that are necessary for autonomous applications, says Mandyam.
Ensuring that the company’s silicon partners are better supported, Arm is evolving its Safety Ready program and is looking to centralise the company’s on-going investment in safety, enabling its silicon partners and the entire automotive supply chain to accelerate individual timelines for bringing safer products to market much faster.
While Arm is looking to integrate the latest certifications and standards in a significant move designed to help the development of autonomous vehicles, it has made available what it says is the first autonomous-class processor with integrated safety, the Cortex-A76AE.
“This processor has been designed for automotive and includes Split-Lock technology, which is available for the first time in application processors and cold be a game changer,” says Mandyam.
The Cortex-A76AE is a CPU that has been uniquely designed for automotive and optimised for 7nm process nodes.
The AE stands for “Automotive Enhanced” and any Arm IP with the AE designator will include specific features addressing the requirements of in-vehicle processing.
“A high level of processing capability is required for autonomous driving, with inherent safety as standard,” explains Mandyam, “and the Cortex-A76AE delivers both. It’s the industry’s first high-performance application processor with Split-Lock capability, combining the processing performance required for autonomous applications and high-integrity safety.”
While Split-Lock is certainly not new to the industry, Arm is the first to introduce it to a processor that has been designed specifically for high performance automotive applications such as autonomous drive.
“Split-Lock delivers the flexibility that’s not currently available in previous lock-step CPU implementations; it means that CPU clusters in an a SoC can be configured either in ‘split mode’ for high performance, where two (or four) independent CPUs in the cluster that can be used for diverse tasks and applications or r ‘lock mode’ where CPUs are in lock-step, creating one (or two) pairs of locked CPUs in a cluster, for higher safety integrity applications,” Mandyam explains.
The CPU clusters can also be configured to operate in a mix of either mode, post silicon production.
Automotive makers can also design their autonomous systems to require watts and not the kilowatts required for today’s prototypes due to the power-efficient computing available in the Cortex-A76AE.
“Lower power also enables a more energy-efficient use of vehicle battery power combined with thermal efficiency to aid the packaging of compute capability while extending the range of vehicles for a lower total cost of driving,” according to Mandyam.
Arm is also introducing new Automotive Enhanced system IP for designing a comprehensive autonomous-class SoC.
The new CoreLink GIC-600AE, CoreLink MMU-600AE and CoreLink CMN-600AE will provide critical elements such as high-performance interrupt management, extended virtualization and memory management, and connectivity to multiple CPU clusters to scale performance in safe multicore systems.
“These products have been designed to enable high-performance systems, targeting ASIL-B to ASIL-D safety integrity, and support the Split-Lock and systematic capabilities for functional safety designed into the Cortex-A76AE.”
The Cortex-A76AE is the first in a roadmap of “Automotive Enhanced” processors which will deliver the fullest functional safety capable IP portfolio in the industry. The new roadmap includes “Helios-AE” and “Hercules-AE”, all optimised for 7nm. More details will be available, as these products are launched.
According to Mandyam, “Arm and its developer ecosystem are simplifying and reducing costs across all layers of automotive software stacks and providing tools on a common architecture.
“Our aim has to be to ensure that safety is not an afterthought and to help car makers earn the consumer trust required for the mass deployment of safe and fully-autonomous vehicles.”
