AIoT startup moves to commercialize AI-in-sensor chips

December 17, 2020 // By Rich Pell
AIoT startup moves to commercialize AI-in-sensor chips
AI-in-sensor processor startup AIStorm says that it has closed $16 million in an oversubscribed Series B financing round to commercialize its AI-in-sensor chips.

The company, which develops chipsets that can process data directly from devices, says the funding will enable it to scale manufacturing to deploy its charge-domain-processing, AI-in-sensor technology in handset, notebook, smart speaker, home automation, remote control, TV, earbud, IoT, wearable, and transport applications.

"AIStorm’s approach is the only technology that allows the sensor to couple directly to popular convolutional neural networks," says David Schie, CEO of AIStorm. "This allows AIStorm to deliver true, template-based 'always-on' at the edge—at prices that deep-submicron digital competitors cannot match for the same performance. This is a winning formula in the highly competitive IoT sector."

Participation in the latest funding round included strategic investors AsusTek, a leading manufacturer of computing equipment; Egis Technology Inc., a major biometrics supplier; Knowles Corporation, a market leader in advanced micro-acoustic microphones; Meyer Corporation, a world leader in food-preparation equipment; and Senvest Management, a New York-based venture capital firm.

The company says that it is targeting a range of applications where its technology offers unique advantages: imaging and audio, facial recognition, access control, behavioral tracking, people counting, face detection, object tracking, and segmentation. The company's charge-domain processing approach differs from solutions used by ARM licensees, FPGAs, or vendors of process-in-memory solutions - instead it accepts charge directly from the sensor, such as electrons from a pixel or MEMs microphone, and multiplies that charge directly.

Users of its technology, says the company, benefit from an AI-at-the-edge technology that is scalable, offers MAC efficiencies up to several thousand TOPS per watt, and works with conventional tool flows.

Jonney Shih, chairman of AsusTek, adds, "AIStorm’s unique, charge-switched analog computing, coupled with its wise AI-in-Sensor approach, offers dramatic advantages for ultra-low power, ultra-low-latency AIoT edge computing. We look forward to an ongoing strategic collaboration."


Related articles:
Intelligent vision sensors have AI processing functionality
Edge AI partnership combines neural sensor SoC, TinyML platform
AI chipsets for edge-vision computing - whitepaper
AI edge processor for

Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.