OmniVision Launches Ultra-Low-Power OAX7700 Vision AI Coprocessor for Vehicles
OmniVision introduced the OAX7700, an ultra-low-power always-on vision AI companion processor equipped with an integrated NPU that reduces parking monitoring power consumption by 97%. It features edge AI capabilities like human detection, object recognition, and distance measurement, enabling continuous monitoring with instantaneous wake-up upon event detection. Continuous parking security traditionally drains vehicle batteries significantly, forcing trade-offs between vehicle protection and battery life. By offloading edge AI processing to a tiny dedicated chip, vehicle makers can provide 24/7 intelligent surveillance without draining the main battery or overloading the central compute platform. The OAX7700 comes in a compact 5.5×3.5 mm 60-pin BGA package with stacked PSRAM and flash memory, supporting feature customization and over-the-air updates. Currently sampling, the chip is targeted for mass production in the first quarter of 2027.
## BACKGROUND
Automotive parking surveillance systems typically rely on high-power central Electronic Control Units (ECUs) to process video feeds continuously, which wastes energy during idle periods. Pseudostatic RAM (PSRAM) is a low-power DRAM variant commonly used in embedded IoT and edge AI devices to deliver high memory bandwidth while minimizing idle power consumption.