Categories: Blog, Industry News

Share

Tech Alert: 5 Ways Edge AI Will Change the World of Technology

BrainChip expects major advancements in industrial and commercial AI devices

AI has spawned a monumental market for both software and hardware and has transformed nearly every industry. The breakthroughs have been extraordinary, and the growth rate of applications has been explosive, yet they represent only the tip of the iceberg for the potential offered by AI.

Today’s edge devices and edge applications have thus far been limited in their capabilities due to the challenges of working around resource constraints, including the availability of limited electrical power, size, internet connectivity, and processing performance. According to experts at BrainChip Inc. (A Delaware company and member of BrainChip Holdings Ltd. ASX: BRN), a leading provider of ultra-low power high performance AI technology, tomorrow’s edge AI will deliver transformational opportunity.

Here are five technologies poised for major change when edge AI is fully realized:

  1. Smart Cameras

Today’s smart camera platforms typically depend on connectivity to the internet and power source, which can be extremely problematic. Video cameras performing various types of object detection and recognition are often mounted in areas in which power and high-speed connectivity are not available. If the network connection is too slow or unreliable, objects move out of frame before they are recognized. Losing connectivity renders these cameras useless. There are also privacy issues associated with images that travel via internet, which can be hacked and exposed.

Tomorrow’s smart cameras will perform both detection, recognition, and learning within the camera itself, without sending images to a back-end system for processing. This is far more efficient and protects data privacy. When the camera itself is embedded with sufficient edge AI processing resources, it can train itself to recognize new images to continually improve its accuracy. As edge AI devices improve, even a solar or battery-powered smart camera will be able to perform facial recognition or object recognition.

  1. Medical diagnostic and safety monitoring

The low power consumption, ability to learn instantly and high speed processing of BrainChip’s edge AI processors makes them suitable to perform fast diagnostics, far beyond the capabilities of today’s smart watches and fitness monitors. Equipped with the right sensors, forthcoming AI processors can learn to identify viruses, such as Covid-19, cancer, influenza, gastroentheristis and other diseases from breath samples just like sniffer dogs can do this today, for instance sampling the breath as you speak into your mobile phone. An edge AI equipped camera can evaluate X-ray images to look for discrepancies or IR based cameras can measure body temperature, and wearable heart monitors can warn of an imminent heart attack and even communicate with a mobile phone to call the hospital.

  1. Advanced Driver Assistance Systems (ADAS)

ADAS is an ideal use case for edge AI because it is an extremely latency-sensitive application, so systems that depend on external connectivity are completely impractical—excess delay or a loss of signal could result in a serious accident. Today, BrainChip’s edge AI processors don’t need remote connectivity, and work in ADAS systems without the risk of unexpected latency or loss of connectivity.

This will not only revolutionize ADAS, it will improve public safety immeasurably. With their incremental learning ability, BrainChip’s edge AI processors will give ADAS the critical ability to adapt to the ever-changing world around the vehicle, and adjust to new conditions far more quickly than today’s devices can. By learning in a way that augments what it already knows, training occurs quickly and happens within the device.

  1. Drones

Unmanned Aircraft Systems (UAS) are a major challenge for integrating edge AI applications. Drones are usually equipped with only a single battery that powers the aircraft’s motors plus onboard computing resources. These computing resources consume power, so they reduce the drone’s flight time. The weight of computing systems also affects flight times, since the motors must work harder (and consume additional power) to lift the increased mass. Drones are prone to latency, interference, and loss of signal, so disruptions can mean catastrophic loss of the aircraft in situations where the system is processing sensor data and feeding it to the drone’s flight control system.

Edge AI processors, such as BrainChip’s Akida, will be a breakthrough for UAS use: they will be far more compact, lightweight, and require minuscule power, so they will have a negligible impact on flight time. Embedded edge AI processing will make critical decisions in real time, eliminating risks from degraded or lost signals. Just as important, if not more so, their data can’t be hacked by someone on the ground.

  1. User Interfaces

The compact size and low power requirements of forthcoming edge AI processors make them ideal for next-generation user interfaces for devices such as a smart speaker that listens for keywords, or a robot that can respond to voices or hand gestures, as well as laptops, tablets, security systems, smart home devices, appliances, and more.

When edge AI devices, including consumer devices, have the ability to learn incrementally, enabled by BrainChip’s Akida processor, users can personalize their devices to recognize their own unique words, voice, commands, or gestures, and increase privacy of their personal data. They can also be used in situations where the availability of electrical power may be a barrier such as outdoors or in inaccessible locations.

  1. Cybersecurity

Edge AI processors will dramatically enhance security of all manner of edge devices and applications. As important as security is, it’s not usually a device’s primary function; rather it’s a feature added on to protect that primary function. However, today, security-related functions are limited to methods that don’t consume excessive resources or constrain the device’s performance of its designated task.

By offloading malware detection and other security processes from a device’s primary CPU to advanced edge AI processors, such as BrainChip’s Akida processor, devices will perform optimally and continuously be protected from cyberthreats. After they learn what normal network traffic patterns look like they will be able to detect malware, attack signatures, and other types of malicious activity. They can quickly learn new attack patterns to adapt to next-generation threats.

The improvements in energy and space efficiency, plus the advances in AI neural processing, will revolutionize edge computing and Internet of Things in ways we are only beginning to grasp.  Today edge devices typically require data center or cloud connectivity, but once that restriction is removed, and they have more processing muscle, we will see enormous changes in industrial and commercial technologies that incorporate AI at the edge.”

BrainChip has developed an advanced neural networking processor that brings artificial intelligence to the edge in a way that existing technologies are not capable. This innovative, event-based, neural network processor is inspired by the event-based nature of the human brain. The resulting technology is high performance, small, ultra-low power and enables a wide array of edge capabilities including inference and incremental learning.

 

Related Posts

View all
  • Linley Fall Processor Conference November 1-2, 2022 Santa Clara, CA (+ Virtual) Please join BrainChip at the upcoming Linley Fall Processor Conference on November 1st and 2nd, 2022 at the Hyatt Regency Hotel, Santa Clara, CA (Virtual attendance option is available) Presentations will address processors and IP cores for AI applications, embedded, data-center, automotive, and server […]

    Continue reading
  • Continue reading
  • Conventional AI silicon and cloud-centric inference models do not perform efficiently at the automotive edge. As many semiconductor companies have already realized, latency and power are two primary issues that must be effectively addressed before the automotive industry can manufacture a new generation of smarter and safer cars. To meet consumer expectations, these vehicles need […]

    Continue reading
  • Join BrainChip at this upcoming Summit. September 14-15, 2022 – Santa Clara, CA The community’s goal is to reduce time-to-value in the ML lifecycle and to unlock new possibilities for AI development. This involves a full-stack effort of efficient operationalization of AI in organizations, productionization of models, tight hw/sw co-design, and best-in-class microarchitectures. The goal […]

    Continue reading