Camera cube reference design delivers AI vision and hearing at the edge

Maxim Integrated Products, Inc. has launched the MAXREFDES178# camera cube reference design that delivers low latency artificial intelligence (AI) vision and hearing inferences for coin cell power budgets. The reference design demonstrates how AI applications previously limited to machines with large power and cost budgets can be embedded in space-constrained, battery-powered edge devices.

(Source: Maxim Integrated Products)

Enabling ultra-low-power internet of things (IoT) devices to implement hearing and vision, the MAXREFDES178# is comprised of the MAX78000 low-power microcontroller with a neural network accelerator for audio and video inferences, the MAX32666 ultra-low power Bluetooth microcontroller, and two MAX9867 audio CODECs.

The MAX78000 solution is up to 50 percent smaller than the next-smallest GPU-based processor, according to the company, and does not require other components like memories or power supplies to implement AI inferences. In addition, the MAX78000’s AI accelerator is said to cut the power of AI inferences up to 1,000× for vision and hearing applications, as compared to other embedded solutions. The AI inferences running on the MAXREFDES178# also show dramatic latency improvements, said Microchip, running more than 100× faster than on an embedded microcontroller.

The system is packaged in an ultra-compact form factor, measuring 1.6 × 1.7 × 1.5 inches (41 × 44 × 39 mm), and demonstrates how AI applications such as facial identification and keyword recognition can be embedded in low-power, cost-sensitive applications such as wearables and IoT devices, said the company.

(Source: Maxim Integrated Products)

Available at Maxim Integrated’s website, the MAXREFDES178#, MAX78000, MAX32666GWPBT+T, and MAX9867EWV+T devices are priced at $149, $8.50, $6.85, and $1.00, respectively, all at 1000-up, FOB USA. All solutions also are available from authorized distributors.