Recapping re:Invent 2020: What’s New in CPG, Part 1
AWS re:Invent 2020 was a huge success! It’s our annual conference for AWS leaders to share the latest advances in AWS technologies. For four weeks, we heard about hundreds of cutting-edge innovations and new product announcements. There were so many new products relevant to the CPG industry, but I’ll only be able cover some of them in this blog and will follow up with a second blog in the coming weeks.
CPGs are turning to AWS to help with their “Make. Move. Market.” transformation of their businesses, in addition to driving business insights. In this blog, I’ll focus on the “Make” technologies that power innovative production processes at scale.
Edge Intelligence – Industrial
re:Invent 2020 introduced a slew of new features and capabilities in the industrial space. CPGs who are also manufacturers can use machine learning (ML) intelligence at the edge without the need for specialized data science expertise.
Amazon Monitron is an end-to-end, ML-based system to detect abnormal behavior in industrial machinery. By taking a predictive, proactive approach to maintenance, you can reduce unplanned downtime. It uses small 50-gram sensors to capture vibration and temperature data from equipment (motors, gearboxes, pumps, etc.) and pre-defined ML algorithms to predict potential failures.
Amazon Lookout for Equipment uses the data from your sensors, such as pressure, flow rate, and RPM, and sophisticated, pre-built ML algorithms to detect abnormal behaviors, so you can act before machine failures occur. You can also set alarms and create automatic trouble tickets when the solution detects an anomaly. Typical equipment maintenance solutions use rule-based methods to detect equipment failures; however, since Amazon Lookout for Equipment uses ML, you can respond faster to prevent costly equipment failures while adopting a data-driven work culture.
Amazon Lookout for Vision can spot industrial product defects accurately at scale. The service uses ML-based computer vision to automate the inspection process to detect missing components, damaged products, irregularities in production lines—all without the need for ML expertise. Dafgårds, Sweden’s largest private food manufacturer, has been using Amazon Lookout for Vision on its frozen pizza production line. The company manufactures 15 different types of frozen pizzas at a rate of more than 100 pizzas per minute. Dafgårds uses Amazon Lookout for Vision to automate the inspection process, ensuring consistent cheese and topping coverage on all of its pizzas.
AWS Panorama is an ML appliance and software development kit (SDK) that enables manufacturers to add computer vision capabilities to on-premises cameras. Since the service provides high accuracy and low latency, you can automate tasks like monitoring workplace safety and security, as well as identify bottlenecks in industrial processes, even in environments with limited or no internet connectivity. Another use case for AWS Panorama is to optimize the loading and unloading of delivery trucks. By adding computer vision capabilities to cameras that monitor the flow of trucks coming into your facility, the solution can assess the size of each truck, so you can proactively route trucks to the optimal loading dock.
If you want more information on AWS solutions for the industrial segment, check out this blog post.
Edge Intelligence – Foundational Services
Amazon EKS Anywhere is a new deployment option for Amazon Elastic Kubernetes Service (Amazon EKS) and enables you to use Kubernetes clusters for on-premises environments, including your own virtual machines and bare metal servers. You can use Amazon EKS Anywhere to train ML models in the cloud while running inference at the edge. It also allows you to create a hybrid architecture, so you can easily migrate on-premises workloads to the cloud. Also check out this blog on Amazon Elastic Container Service (Amazon ECS) Anywhere for additional information.
Amazon SageMaker Edge Manager allows you to manage ML models across a fleet of edge devices, such as smart cameras, robots, and mobile devices. It optimizes ML models to run faster on target devices, and it provides dashboards to track and analyze ML models. Also check out Amazon SageMaker Neo.
For smaller locations, AWS Outposts 1U and 2U form factors are rack-mountable servers designed specifically for low-latency or local data processing in on-premises locations, like factory floors. They can be used as a manufacturing process control system that responds in near real-time to factory floor equipment.
Building on Amazon’s Consumer Business Experience
Many of the new edge services I covered use ML, which relies on two decades of learning from Amazon’s own experience, validating the power of these products. For example: Amazon Monitron uses the same technology as Amazon Fulfillments Centers to monitor equipment, and Amazon Lookout for Metrics uses ML models informed by 20+ years of Amazon experience to detect anomalies in business and operational data. I’ll have more to say about this in my next blog.
These new services provide a unique, differentiated value to our CPG customers, leveraging our deep expertise and understanding of this massive market segment. To learn more, visit AWS for CPG or get your questions answered here.