Arm Enables Next-Generation Autonomous Systems

Autonomy gives computers the power to compute and make decisions over workloads with little to no human input. Autonomous systems represent a shift change in technological design, enabling a vast range of new use cases with the potential to transform the way we live, work, move and communicate.

Arm’s dedicated suite of autonomous IP, including processor (CPU), graphics processor (GPU) and image signal processor (ISP), is designed from the ground up to provide autonomous developers with everything they need to design and deploy next-generation autonomous systems.

What defines an autonomous system?

Since early humans learned to fashion tools from rock, technological evolution has been driven by desire to automate processes that would otherwise be too difficult or time-consuming. The crafted edge of a stone axe enables the felling of a tree for firewood before nightfall. A pocket calculator solves a math problem far more quickly than even the greatest mind could. A data center can spot patterns in a database of thousands of images almost instantaneously.

However, much of the technology that exists today still relies upon human involvement at both the input stage and in making decisions over the output. And while technology has evolved to do its part of the process faster and more efficiently, humans have not. In short, we humans have become the bottleneck— and the solution is to give compute systems the power of informed, autonomous decision making.

Autonomous decision making

Autonomous systems already exist in a growing number of emerging use cases and in devices we use every day. They’re in our smartphones, when our device unlocks itself having decided that our face meets the visual criteria. They’re increasingly in our cars: while the fully autonomous vehicle or ‘self-driving car’ may still be some way away from commercial deployment, automation features built into Advanced Driver Assistance Systems (ADAS) are already helping to reduce the number of accidents by up to 40 percent.

As 94 percent of road traffic accidents occur due to human error, giving a computer some form of control—such as maintaining the vehicle’s position within a lane on the highway—simply makes sense. We are also seeing autonomous systems that are more limited in scope than fully self-driving cars but still have huge transformative potential, such as last-mile delivery and fixed-route shuttles. These kinds of services are a near-term deployment reality.

Autonomous systems also increasingly prevalent within industry, particularly within smart manufacturing environments. Like with cars, we’re some way away from a full ‘lights out’ scenario in which autonomous robots perform all tasks within a factory. But we are already seeing deployment of robots operating autonomously within fixed physical boundaries within warehouses, and machine learning (ML) used to identify issues with items on a production line or take systems offline if they are deemed to be showing signs of imminent failure.

Autonomous systems are also already enabling aspects of Smart Manufacturing and Warehousing, optimizing the production line and associated logistics through more efficient use of resources and the ability to meet changing customer and logistical needs.

Key enabling technologies needed for next-generation autonomous systems

These examples represent the beginning of a journey towards true technological autonomy. But it’s a journey that will not occur overnight, and one that will require key enabling technologies along the way.

These enabling technologies must meet certain requirements to ensure systems are:

• Conformant to relevant safety standards to enable building a safe system
• Scalable in order to address a range of workloads and processing capabilities
• Able to operate, react and respond in real-time
• Secure in their operation
• Energy-efficient in their operation
• Able to understand their surroundings using sensors such as cameras
• Accessible for developers of autonomous applications

Arm technology addresses the needs of the ‘next mile’ in autonomous systems design

Arm’s suite of autonomous IP is designed to make it easier than ever before for developers of future autonomous systems to bring their vision to life as quickly, securely and efficiently as possible, while addressing industry safety standards such as ISO 26262 for automotive and IEC 61508 for industrial.

Arm Cortex-A78AE is Arm’s highest performance safety capable processor (CPU), delivering 30 percent higher performance than its predecessor at improved power efficiency, and is well suited to automated driving and industrial robotics applications. Cortex-A78AE supports features to achieve key Safety Integrity Level (SIL) and Automotive Safety Integrity Level (ASIL) risk classifications as well as building on the capabilities of the Split-Lock safety architecture introduced in 2018 with a new Hybrid mode that improves performance in certain safety scenarios.

Arm Mali-G78AE is Arm’s first safety-capable graphics processor (GPU), including a new Flexible Partitioning feature which allows GPU resources to be allocated to independent hardware partitions to separate autonomous workloads from each other, enhancing safety. Mali-G78AE includes features to help meet ASIL B / SIL 2 requirements and is ideal for applications including safety related user interfaces in cars or path planning in industrial robots.

Arm Mali-C71AE is an image signal processor (ISP) that ensures maximum information is extracted from multi-camera autonomous vision systems, which may need to operate in a variety of lighting conditions. Mali-C71AE includes safety features to help achieve ASIL B / SIL 2 level of integrity and address vision applications across automotive and industrial applications.

Integrating these three processing technologies into a system on chip (SoC) will provide the power-efficient and safety-enabled processing needed to unlock the decision-making potential of autonomous systems.

Beyond these new amazing hardware technologies, we are working to enable the developers of autonomous systems. For software developers, we are enabling familiar cloud-native technologies in autonomous applications to ease development, while Arm development solutions accelerate software development and validation while shortening the path to deployment. For developers of autonomous silicon, our physical IP, training and design reviews help reduce risk.

Through empowering next-generation autonomous systems to make independent decisions and safely complete tasks on their own, Arm enables a world of new possibilities for autonomous developers. We believe that autonomy and autonomous thinking will become the predominant computing workload of the future. Developers deploying Arm autonomous IP stand to enable the flexible, scalable heterogeneous compute required to power next-generation autonomous systems and applications.