Sensory Expands ARM Support For Embedded AI

May 26, 2017 - Written By Daniel Fuller

Sensory Inc has announced an expansion of its efforts to bring embedded machine learning, neural networking, and AI to ARM devices. Sensory’s technologies allow on-device AI operations in devices that wouldn’t normally support them, including many ARM products. The company’s offerings on ARM platforms now include deep neural networking, always-listening wake word support, speech recognition, chat bots, computer vision, voice biometric applications, and statistic language models, Sensory said. The new programs are supported on virtually all ARM-based processors, though some of them will require significant amounts of power to run. These advancements will allow devices like Internet of Things (IoT) hubs, smartphones, in-car entertainment systems, and even items built with design boards like the Raspberry Pi to run a wide range of powerful AI applications.

The press release from Sensory did not mention Qualcomm’s Snapdragon 835 processor, but the implications are quite obvious. This flagship system-on-chip already supports machine learning through native support for Google’s TensorFlow and multi-device neural networking. Adding in Sensory’s suite of boosts, tweaks, and pre-made applications will make it easier for OS and application developers to take full advantage of the processor’s capabilities. Full speech recognition with training, simulated neural networking, and adaptive, conversational AI are just a few things that this processor can do directly on a device, with or without a network connection. Sensory’s expansion on ARM platforms brings new pre-built applications and administration methods to the SoC, giving developers a base to work on, and more flexibility in creating machine learning applications.

Sensory Inc is not a new name in the industry; the Silicon Valley company has been working on pushing machine learning and AI applications closer to its customers for years now, with a number of prominent clients to show for it. Samsung has tapped Sensory in the past for technology built into the Galaxy phone lineups, and its other renowned clients include Huawei, Lenovo, AT&T, LG, Honeywell, and even toy makers Fisher Price and Hasbro. Likewise, Amazon uses Sensory’s technology as a cornerstone in the widely used Alexa voice and AI platform, as seen on the Amazon Echo. An update on the company’s endeavors will likely follow later this year.