Embedded World 2024: Qualcomm takes on-device AI to mid-tier IoT

At the beginning of 2024, mobile process and platform giant Qualcomm Technologies set out its stall for the year regarding how it would use on-device artificial intelligence (AI) to drive success in a number of use cases – in particular automotive and industry 4.0 applications – and in the latest part of this drive, the company has launched industrial and embedded AI platforms, as well as micro-power Wi-Fi silicon on chip (SoC) to enable intelligent computing everywhere.

Taking advantage of Embedded World 2024, Qualcomm Technologies has unveiled new embedded ecosystem products in the form of the QCC730 Wi-Fi service and RB3 Gen 2 Platform to provide what it calls critical upgrades to enable on-device AI, high-performance, low-power processing and connectivity for internet of things (IoT) products and applications.

QCC730 is described as a “disruptive” micro-power Wi-Fi system for IoT connectivity, using up to 88% less power than previous generations – and, claims Qualcomm, can no less than “revolutionise” products in battery-powered industrial, commercial and consumer applications. QCC730 will be complemented with an open-source IDE and SDK that supports cloud connectivity offloading for ease of development.

Developers can also implement QCC730 as an alternative to Bluetooth IoT applications for flexible design and direct cloud connectivity. Furthermore, Qualcomm offers a family of IoT connectivity products including QCC711, a tri-core Bluetooth Low Energy SoC and QCC740, an all-in-one service supporting Thread, Zigbee, Wi-Fi and Bluetooth.

“Complementing high-performance, low-latency wireless connectivity solutions, Qualcomm QCC730 SoC is an industry-leading micro-powered Wi-Fi solution enabling Wi-Fi for the world of battery-powered IoT platforms,” said Rahul Patel, group general manager of connectivity, broadband and networking at Qualcomm Technologies. “QCC730 enables devices to support TCP/IP networking capabilities while remaining form-factor and complete wireless constrained, whilst remaining connected to the Cloud platforms.”

Based on the Qualcomm QCS6490 processor, the RB3 Gen 2 Platform hardware and software offering is designed for IoT and embedded applications. It is attributed with offering a combination of high-performance processing, 10x increase in on-device AI processing, support for quadruple 8MP+ camera sensors, computer vision and integrated Wi-Fi 6E.

Qualcomm expects RB3 Gen 2 to see use in a wide range of products, including various types of robots, drones, industrial handheld devices, industrial and connected cameras, AI edge boxes, and intelligent displays.

The platform is now available for pre-order in two integrated development kits, and supports downloadable software updates to simplify application development and integration, and build proof-of-concepts and prototypes.

The RB3 Gen 2 is also supported in the recently announced Qualcomm AI Hub, which contains a library of continuously refreshed pre-optimised AI models. This is said to result in superior on-device AI performance, lower memory utilisation, and power optimised operation. This allows for an out-of-the-box optimised experience across broadly used AI models deployed across IoT and embedded applications.

Developers can view a selection of models for RB3 Gen 2 and integrate the optimised AI models into their applications to reduce time-to-market and unlock the benefits of on-device AI implementations such as immediacy, reliability, privacy, personalisation and cost savings.

To expand on its portfolio of IoT services, Qualcomm Technologies also announced it will introduce an industrial-grade platform that will focus on addressing the functional safety, and environmental and mechanical handling requirements in industrial applications.

This platform will support System Integrity Level certification, wide operating temperature ranges and industrial module packaging to address the deployment requirements in enterprise and industrial environments. This offering is expected to be available in June 2024, and will feature high-performance central processing unit, graphics processing unit and on-device AI capabilities, advanced safe camera ISP with multi-concurrent camera support, and support for industrial input/output needs.