IoT in Action: How Edge AI is reshaping retail

The retail landscape is undergoing a drastic transformation. As technology advances, IoT and artificial intelligence (AI) solutions are becoming increasingly affordable and will soon become ubiquitous across the industry.

That’s because when combined with AI, IoT enables retailers to connect the physical and digital worlds. According to the IoT signals report, the top IoT scenarios for retailers include security and surveillance, inventory management, and supply chain optimization. In fact, 70% of retailers plan to invest in IoT by 2021.[i]

Of course, there are plenty of other usage scenarios as well. By combining AI and IoT, retailers can track and optimize shopper traffic, personalize customer interactions, and right-size customer service levels. From an operational standpoint, retailers are leveraging IoT to keep shelves stocked, reduce product shrinkage, and lower energy costs. The possibilities are nearly limitless.

However, in order to implement IoT solutions, retailers must overcome barriers such as complexity, time to value, and scalability. Add to that data sovereignty and privacy considerations along with manageability and long-term flexibility, getting a solution up and running can prove challenging.

Below we’ll explore some key trends that are helping retailers push past obstacles and combine IoT, AI, and edge technology to transform business. For more, be sure to check out the Microsoft AI@Edge community and register for IoT in Action in New York.

AI at the edge

AI is moving from the cloud to the edge. By shifting certain workloads to the edge of the network, edge devices can run AI algorithms to analyze and act on data locally and send only what’s needed to the cloud for further analysis. In addition to reducing bandwidth, AI at the edge facilitates real-time decision making. It can even enable offline operations, whereby devices transmit data to the cloud when connected and process and act on data when not.

An organization’s data is one of its key assets, and data privacy and security must be a key priority. AI at the edge can help mitigate these concerns. As an example, let’s say you are running a series of IoT-connected 4K security cameras snapping 60 frames per minute. In addition to eating up expensive bandwidth, uploading customer images can create privacy concerns. But thanks to edge computing, these images can be automatically blurred or analyzed locally, sending only what’s needed to the cloud.

Vision AI for retail

And this brings us to vision AI. In the next ten years, perception AI (including vision AI) will become as commonplace as silicon and software. Vision AI offers powerful benefits, especially when combined with the power of AI and edge computing. New hardware and technical innovations allow for faster inferencing from multiple camera streams, giving retailers the ability to run multiple serialized AI models.

Vision AI also enables retailers to monitor product levels to keep shelves stocked. It can analyze customers’ in-store movements and behaviors to optimize product placement and store layout as well as minimize wait times at checkout. Vision AI can even watch for safety hazards and send alerts when, for instance, a product has spilled.

IoT-enabling existing assets

While IoT has become more affordable than ever, retailers still face challenges getting solutions off the ground. We have found that retailers have the best chance for success by starting small, using experimentation and proof of concepts to clarify value and ROI.

From a cost standpoint, it’s not always necessary to replace equipment. Increasingly retailers are turning to brownfield installations—retrofitting existing equipment with IoT sensors—to save money and accelerate ROI. This is true especially for some vision AI related scenarios where existing cameras can be plugged into smart gateways that run AI models and make them intelligent.

Starbucks was recently in the news for IoT-connecting its existing coffee machines using the new guardian module from Microsoft. The guardian module brings Azure Sphere security to brownfield IoT installations, facilitating secured operations between the device and the cloud. By collecting data points such as coffee temperature and water quality, Starbucks can proactively identify and mitigate maintenance issues and avoid costly downtime.

Faster provisioning and deployments

Microsoft’s Azure IoT solutions continue to advance. Recent product announcements have focused on reduced complexity, repeatable solutions, and easier device provisioning at scale. Furthermore, the introduction of open formats makes it possible for developers to create highly secure, extensible solutions that can be modified in the future for long-term success.

Register for IoT in Action and join the Microsoft AI@Edge community

Register for an IoT in Action event, and discover how IoT technology platforms are helping retailers drive shopping analytics, smart customer experiences, and store efficiencies. You’ll learn about the latest trends and future direction of IoT solutions in retail. You’ll also connect with the partners that can help you combine AI, edge, and cloud technologies to create a secure, sovereign, and flexible IoT solution that will position you for success for years to come.

Can’t make it to the events? Check out our collection of IoT in Action webinars. In addition, be sure to join the Microsoft AI@Edge community which bundles together hardware, machine learning, and Azure services to help you start building solutions for the intelligent edge. Explore a variety of projects and get the resources you need to begin your proof of concept.

[i] Zebra Technologies, “2017 Retail Vision Study,” found at: https://www.zebra.com/content/dam/zebra_new_ia/en-us/solutions-verticals/vertical-solutions/retail/vision-study/retail-vision-study-2017-en-gb.pdf, accessed Dec. 18, 2019