Qualcomm is putting its processing muscle behind a new Internet of Things (IoT) platform called Vision Intelligence. What makes this family of system-on-chips (SoC) special, according to the chipset giant, is that it uses an artificial intelligence (AI) engine (coupled with a heterogeneous compute architecture and an ARM-based CPU) to make it possible to do processing at the network edge instead of in the cloud.
“This is what customers have been asking for,” said Joseph Bousaba, Qualcomm’s vice president of product management. “They want advanced capabilities like AI or machine learning, but they want it at the edge. Because of the low latency and the responsiveness, the device is much faster.”
Security is also enhanced, he said. Because when data is sent from the device to the cloud and back there are more potential vulnerabilities. “When you access the internet and go into the cloud, there are vulnerabilities to that. Security is more of an issue.”
Vision Intelligence is Qualcomm’s flagship platform for IoT applications. Bousaba said that the reason it’s taken awhile to get a SoC platform to do edge processing is because this type of compute engine hasn’t existed until now. “We are able to get the processing capabilities in the device itself,” he said.
The QCS605 and QCS603 SoCs have Qualcomm’s image signal processor as well as its AI engine. Qualcomm’s AI Engine incorporates the company’s Snapdragon Neural Processing Engine software framework.
In addition, the SoCs are equipped with machine learning and computer vision software development kits. Bousaba said these capabilities will help cameras in devices handle additional tasks like tracking, object detection, classification, and facial recognition.
Qualcomm said that the SoCs are sampling now and Kedacom, which makes video surveillance equipment, and Ricoh Theta, which makes cameras, have already said they will develop products based upon the Vision Intelligence platform. The company expects that products based upon this IoT platform will be available in the second half of the year.
Bousaba said that Qualcomm plans to add more capabilities around machine learning and performance optimization in the future.
Qualcomm’s Edge Journey
This isn’t the first time Qualcomm has talked about the growing demand for edge computing devices. Earlier this year the company introduced an edge software designed to help IoT customers provision, connect, and manage all their IoT devices through their cloud platforms.
At the time, Qualcomm said it was looking to improve three areas in IoT: device security, lifecycle management of IoT devices, and closer collaboration between cloud providers and edge devices.