Eyeris DNNs are carefully designed for functional safety standards, flexible camera placements, and efficient inference on low-power edge processors.
Fremont, CA: Eyeris Technologies, a world leader in vision AI for in-cabin sensing, announced its collaboration with Texas Instruments (TI) on an industry-first in-cabin sensing AI solution, using TI's Jacinto TDA4 processors and 2D RGB-IR image sensors, at the Consumer Electronics Show (CES) 2021. D3 Engineering designs products and integrates technologies from TI and Eyeris using its DesignCore platform to show this advanced in-cabin monitoring solution for autonomous and highly automated vehicles. This fully integrated in-cabin sensing AI solution will be demonstratable to select customers.
The TDA4 SoC processor enables efficient inference of the Eyeris deep neural network (DNN) portfolio from multiple RGB-IR cameras simultaneously. This complete in-cabin monitoring solution, integrated into the D3 Engineering DesignCore platform, is designed for rugged, field-deployable, and production-intent units. Also, RGB-IR image sensors are quickly becoming an industry requirement because of their unique ability to accurately capture the in-cabin space under the widest range of lighting conditions.
Eyeris in-cabin sensing portfolio of AI algorithms includes a driver monitoring system (DMS) and occupant monitoring system (OMS) features that adhere to global NCAP requirements.
"As we continue to expand our hardware partnerships, teaming up with a well-established automotive processor company provides our automotive OEM customers with another compelling option for mass production," stated Modar Alaoui, founder and chief executive officer of Eyeris. "Collaborating with TI is another step forward where Eyeris continues to raise the bar for a complete in-cabin sensing solution. We are also delighted that D3 Engineering is providing our joint technology on their DesignCore platform."
TI's TDA4VM processors include embedded analytics acceleration to enable efficient deep learning performance within a low-power envelope. Additionally, these processors can perform safety-critical operations up to an ASIL D compliant level while also performing non-safety critical operations on one chip and function at an automotive operating junction temperature of 125 degrees Celsius. The SoC has built-in ISP and vision processing accelerators that help achieve more efficient system performance and reduce development costs.