In the ever-evolving landscape of Artificial Intelligence (AI) integration, Canada-based startup Applied Brain Research (ABR) develops technology that mimics how the brain processes information efficiently.
ABR's primary focus is to enable the integration of advanced AI capabilities into edge IoT devices, enhancing their functionality, responsiveness, and battery life.
Three years ago, ABR's focus was primarily on neuromorphic AI, catering to system integrators through service contracts aimed at software development. However, a significant transformation has since taken place. Today, ABR's mission revolves around enabling the seamless integration of cutting-edge AI capabilities into edge IoT devices.
"Our mission is to enable the integration of leading-edge AI capabilities into edge IoT devices specifically, and that enhance user experience through increasing functionality, responsiveness, and battery life for those products," explained Kevin Conley, the CEO of ABR.
Credit: ABR
To achieve this goal, ABR is developing a Time Series Processing (TSP) ASIC paired with a software stack that facilitates model training, compilation, and deployment, to provide a complete solution to the customer.
A notable milestone in ABR's journey is the foray into the semiconductor field.
At the heart of ABR's technology lies the Legendre Memory Unit (LMU), a patented network algorithm inspired by how the brain processes time series data. The LMU is the first of a new class of state space neural networks that is mathematically optimized for time series processing efficiency, enabling ABR's TSP ASIC to achieve functionality that would require orders of magnitude more power on competing solutions.
In a crowded landscape of AI chip manufacturers, ABR stands out with its unique approach to time series processing. While competitors focus on optimizing hardware elements or cloud-based solutions, ABR prioritizes algorithmic efficiency tailored for edge devices.
Credit: ABR
By integrating ABR's chip solutions into edge devices, several key advantages are realized. Devices can function without a constant connection to other devices, reducing latency and enabling more accurate inference from larger data sets. Additionally, data privacy is maintained as information does not need to leave the device.
ABR's technology has a wide range of applications, including consumer, medical, industrial, and automotive sectors. However, the company's current focus is on wearables, biomedical devices, and speech recognition applications.
"The most meaningful applications for us are those that realize better health outcomes or simplify user interactions with devices to bring more capability and make it easier for people to benefit from them," Conley stated.
One of ABR's key differentiators is its focus on low-power AI solutions for edge devices. As Conley explained, "We're unique in the fact that we're just focusing there. We don't look at the data center applications, even though ABR's technology applies to a broad set of those applications as well. We're just focused on battery-powered edge AIoT, whether that works with or without connection to mobile phones, tablets, computers, servers, etc."
Looking ahead, ABR aims to expand its technology's reach to enable complete dialogue systems built into devices, revolutionizing human-machine interaction. The company also sees significant potential in automotive applications, such as voice interaction, autonomous vehicle data processing, and battery management systems. "And ultimately, by reducing the power, we're also reducing the carbon footprint of these devices deployed by the billions."
With its brain-inspired AI solutions, ABR is poised to disrupt the edge computing market, delivering unprecedented power efficiency, accuracy, and functionality to a broad range of IoT devices.
