embedUR systems has developed an Ultra-Wideband sensing Edge AI solution using NXP Semiconductor's Trimension NCJ29D6 platform that enables gesture-based vehicle control and advanced automotive sensing applications.
The technology allows drivers to perform touch-free interactions such as opening car trunks with hand or foot gestures and controlling parking assistance functions. The solution operates on a compact 215 KB AI model optimized for real-time gesture detection on NXP's integrated UWB transceiver chip.
NXP's Trimension NCJ29D6 serves as a fully integrated, low-energy UWB transceiver designed for secure ranging and radar applications in automotive environments. The chip supports Digital Key functionality for hands-free vehicle access while enabling radar-based applications including Child Presence Detection and kick-sensing for trunk access.
embedUR systems optimized their UWB Gesture and Motion Recognition Model specifically for the NXP platform, creating a lightweight solution that processes gestures in real-time while maintaining power efficiency on edge devices.
The partnership addresses growing demand for contactless vehicle interactions and advanced sensing capabilities in the automotive sector. Beyond automotive applications, the technology can be adapted for consumer electronics, industrial automation, and smart home systems.
"With our expertise in embedded AI and UWB technology, we are delivering practical, hands-free interaction solutions that enhance safety and convenience," said Eric Smiley, VP of Business Development at embedUR systems. "This project demonstrates how UWB, combined with optimized Edge AI models, can help create advanced sensing applications for automotive and beyond."
Michael Leitner, General Manager, Secure Car Access at NXP Semiconductors, confirms: "The knowledge of embedUR in wireless sensing and efficient implementation of data processing on edge devices positions embedUR to fully leverage the capabilities of our Ultra-Wideband (UWB) devices. EmbedUR demonstrated very positive results in UWB-based gesture sensing and we are looking forward to further applications that embedUR is able to address based on our UWB IC."
embedUR systems brings over two decades of experience in embedded systems and wireless technologies to the partnership. The company has previously developed Edge AI applications including real-time image segmentation models running on Arm Cortex-M7 processors under 300 KB and facial recognition pipelines on low-power edge devices with less than 1MB memory footprints.
The company routinely optimizes AI models to operate within 200-500 KB memory constraints while maintaining accuracy for real-time inference on power-efficient platforms. This expertise in model compression and quantization enables customers to accelerate development cycles and meet strict power and memory requirements.
embedUR systems operates as a Silicon Valley-based leader in embedded software, IoT, and Edge AI solutions, with code running in millions of connected devices worldwide. The company created ModelNova, a resource hub for rapid prototyping and Edge AI deployment designed to support developers building intelligent products.