Chip developer OMNIVISION has announced an AI-enabled application-specific integrated circuit (ASIC) for dedicated driver and occupant monitoring systems (DMS/OMS). It uses a stacked-die architecture to provide integrated RGB-IR image signal processing (ISP) with two AI neural processing units (NPUs) and embedded DDR3 memory (2Gb) in a single low power, small package chip.

The OAX4600 uses OMNIVISION’s dedicated NPU with 2 tera operations per second (TOPS). To give developers greater flexibility, Seeing Machine’s Occula NPU is also included. The OAX4600 features 2Gb of embedded DDR stacked memory and utilises the tri-ARM Cortex A53 with NEON core subsystem.

The OAX4600 is capable of higher resolution processing of up to 5 megapixels. It has fast boot-up time to eliminate any delay between ignition and activation of the interior camera and features integrated cybersecurity.

“Driver monitoring solutions that are high performance and low power, with a small package size, are attractive to OEMs looking to comply to regional safety regulations. The flexibility to add additional features and expand into cabin monitoring is a bonus,” said Mark Fitzgerald, director, Autonomous Vehicle Research, Strategy Analytics.

“Our OAX4600 integrates a unique set of features developed specifically for interior monitoring systems (IMS), and its small size and low power provide OEMs with the flexibility to place the camera at any location within the car, regardless of space constraints,” said Brian Pluckebaum, automotive product marketing manager at OMNIVISION. “It uses OMNIVISION’s ISP processing technology and dedicated NPU to ensure robust, latency-free image processing, even in the case of RGB-IR. We have also designed this device from the ground up with ASIL B functional safety in mind.”