Living on the edge: How edge AI enhances everyday experiences

We use different products in order to live our lives: air conditioning to cool our houses; cars to get from place to place; or solar panels to generate energy. But even though these products are integral to our lives, what if we could improve the user experience so the products are not just a part of, but enhance, our lives? Your HVAC system could be smart enough to tell you when the filter needs to be changed before you can, enabling your air conditioner to run efficiently, keeping you cool throughout the summer.

Applications use neural network-based edge artificial intelligence (AI) to locally understand and improve products, and improve user experience at the place where an event occurs – called the edge. Engineers can design and use data to train neural network algorithms, then execute these algorithms on embedded devices to solve problems. Incorporating edge AI into the applications we use daily will enhance our lives, making the products we encounter everyday easier to use, safer and more sustainable.

Today, when many people think of “smart” devices, they commonly think of wireless connectivity and storing data in the cloud for decision-making. But connected devices aren’t necessarily intelligent. Edge AI brings computational power closer to the data source, enabling decisions to be made on the periphery in the device instead of in the cloud. This allows for lower latency, higher power efficiency and improved robustness and data security, translating to faster decision-making.

Think of a smart home security camera. If you wanted to detect whether an object in your back yard is a cat or a stranger, using cloud AI would require sending the data up to the cloud, processing it and then sending it back down to make the decision. This would not only take too long, but require a lot more power to accomplish. When making decisions at the edge, highly-integrated embedded devices run neural networks locally, which requires less power and allows for improved security and privacy.

Using edge AI to future-proof

Edge AI has the potential to transform how we interact with electronics, making them more responsive, efficient and secure. Running AI algorithms locally, close to where data is collected, allows for faster decision-making. Because of this, decisions made at the edge can potentially save lives. Now, electronics don’t need cloud-based resources to react locally. When you’re driving, vision processors work with radar detection technology to continuously monitor your surroundings, helping your car make quick system decisions in the event of obstacles, or for increased autonomy.

In factories, edge AI can enable motor fault detection. Capturing a fault in the motor before it causes damage to the entire system, and in turn identify patterns for predictive maintenance, can lead to more reliable, efficient and cost-effective operations.

As edge AI capabilities evolve and applications with it become smarter and more connected, it will better adapt to human lifestyles. For example, an air conditioner could use radar to detect where you are in your home and optimize airflow to your location, making your house more energy-efficient. Or radar sensing could turn the television on and off when someone enters your living room.

With edge AI making decisions faster and more efficiently, we’re empowered to strategically use what resources we have and make smarter decisions. Solar panels with edge AI have improved safety with fault detection, detecting power surges and shutting down a system before it ignites, leading to the proliferation of renewable energy sources. Edge AI can complement our decision-making to keep us safe and increase efficiency of the systems we use, while also propelling us to a more sustainable future.

Resources for increased intelligence

As we begin to see edge AI in more applications, I want adopting AI to be easy and accessible for all engineers, regardless of their expertise or experience with AI. In my job, I get to lead teams who equip our customers with the software and hardware tools they need. Building on our company’s commitment to open source and easy-to-use software, we can empower designers to use their data and add meaningful intelligence to their systems.

We believe that semiconductors can improve the world by making electronics more affordable. This passion is nothing new. Our nearly 50 years of digital signal processing technology started as executing efficient mathematical functions with voice and sound, and has enabled us to solve large problems that continue to grow in complexity. Because of our experience in signal processing, we have grown our portfolio to enable innovations with scalable hardware devices for edge AI applications, regardless of the processing power that they need. We have also increased the level of integration in embedded processors to make them more affordable. Our customers can adopt edge AI easily for truly intelligent applications. All of our edge AI contributions have enabled fast and efficient decision-making that has made the world safer and more sustainable. 

The adoption of edge AI in applications will only increase as technology advances. Over the last few years, we have seen systems become increasingly sophisticated, and we’re helping to drive that. I expect that edge AI will illuminate new ways to complement our lives and make decision-making easier.

Given where we were 50 years ago and where we are now, I’m excited to see how we can leverage today’s innovations for tomorrow’s edge AI.

Amichai Ron, senior vice president of embedded processing at our company, authored this blog.