The solution integrates Socionext's compact 60 GHz FMCW radar, capable of capturing detailed 3D environmental information in any light or weather condition, with Innatera's ultra-low-power Spike Neural Processor, which interprets radar patterns directly at the sensor's edge.
The neuromorphic processor distinguishes human from non-human movement—including stationary people—achieving over 99% detection accuracy while maintaining power consumption below one milliwatt. This allows devices like cameras and radios to remain in standby mode until a person's presence is confirmed, extending battery life by three to six times.
“Our 60 GHz FMCW radar offers accurate and privacy-friendly detection in a compact form factor, ideal for battery-powered IoT devices,” said Matthias Neumann, Senior Marketing Manager Smart Devices & Industrial at Socionext. “Together with Innatera, we demonstrated a presence detection solution that raises the bar in accuracy, integration flexibility, and reliability in real-world environments.”.
“Neuromorphic computing is unlocking a new generation of always-on features in devices, thanks to powerful brain-inspired processing with minimal energy consumption ,” said Sumeet Kumar, CEO of Innatera.
“The combination of our Spiking Neural Processor with Socionext’s FMCW radar enables robust, real-time detection of human presence with power levels that no conventional AI system can match.”.
