Retailers buy in to AI at the store’s edge
In the wake of the National Retail Federation’s annual expo, it is clear that edge artificial intelligence is an attractive market for hardware and software makers. Companies ranging from internationally recognized vendors such as Toshiba to tech vendors such as Socionext, Shekel Scales and Cradlepoint were on hand to show how AI is helping boost sales efficiency.
Toshiba Global Commerce Solutions was on hand, pushing its version of frictionless-store technology. The “frictionless store” is an industry concept that describes turning retail shelves into real-time extensions of edge and cloud data and decision software.
The products gather data with Toshiba sensors and machine vision software and analyze it with edge computing for rapid response to help predict events that can reduce store efficiency or result directly in lost revenue.
Toshiba executives are pushing a slate of hardware and software products that range from the very tactical (recognizing items at self-checkout and preventing hardware failures) to the strategic (such as gleaning actionable insights from a never-ending flow of data).
A smaller edge AI player, Shekel Scales, is also targeting the retail industry. Shekel showed up at the retail expo with Visual Recognition, software, and hardware designed to cut errors and time spent by customers at checkout.
Visual Recognition, a computer vision-enabled scale, uses cloudless and serverless self-training AI software from Edgify. It is described as plug-and-play, needing no additional infrastructure.
The system is compatible with the Linux Foundation’s EdgeX open software framework designed to encourage interoperability between Internet of Things (IoT) devices and edge hardware and software.
Socionext Inc. also took advantage of the focus on retail, touting a new retail-ready edge AI server called the Boxiedge that it developed with Foxconn Technology Group and Network Optix Inc. Socionext makes systems-on-a-chip (SoC) hardware. Foxconn is a global player in advanced manufacturing, and Network Optix makes video-management software.
Their fanless server boasts a performance rate of more than 20 tera operations per second (TOPS) while typically consuming 30 watts, according to Socionext. Boxiedge is based on a scalable 24-core ARM Cortex-A53 SoC, and it ships with an AI-accelerating card for object classification. The partners claim the Boxiedge can support demanding edge-computing, IoT and real-time data processing needs while minimizing energy use.
Another firm, Cradlepoint Inc., attended the expo to announce new machine-learning capabilities for its 4G LTE routers and ruggedized machine-to-machine and IoT gateways. The edge hardware is used in large, distributed wireless wide-area networks, which are common in most large retail companies.
The addition is part of Cradlepoint’s NetCloud line of software and services. Executives say the new machine-learning capabilities can “virtually eliminate the risk of (data plan) overages” by monitoring traffic the hardware is handling.
That is a significant selling point for large retailers who must plan and monitor cellular data consumption as they try to stay in front of consumers with endlessly evolving media options.
Article Topics
AI | Cradlepoint | edge computing | image recognition | IoT | retail | Toshiba
Comments