Passionate about building technologies, businesses and organizations that progress our societies forward.View the profile
About the talk
This session highlights how Google Cloud partners closely with manufacturing, industrial, and transportation organizations to drive business transformation. Hear about customer stories as well as Google Cloud’s differentiated AI products and solutions.
Speaker: Mandeep Waraich
Google Cloud Next ’20: OnAir → https://goo.gle/next2020
Subscribe to the GCP Channel → https://goo.gle/GCP
product: Cloud AI & Industry Solutions; fullname: Mandeep Waraich;
event: Google Cloud Next 2020; re_ty: Publish;
Hello everyone and welcome. My name is Monday burright and I'm the head of product Industrial in Google Cloud business. Portfolio for official intelligence products that help solve some of the most complex and hard industrial problems at global scale. Talk about three things. First formation and Industrial efficiency, and how artificial intelligence is revealing an entirely new universe of possibilities. Second, Google clouds approach to Bringing these are
Technologies to the market and third pack of AI at mobile Enterprise scale. Let me first begin by sharing how we are approaching our strategy to help industrial Enterprise customers transform themselves. Hinges on three pillars to provide tools and Technologies to connect your physical assets and boat tours to the cloud. Second is to bring both your existing Alaric stools and new analytics capabilities that augments human decision-making. And started to drive into and automation of industrial processes by embedding a different shade idiotic knowledge, he's into the industrial work
clothes. As far as the session I'll book is only on our AI capabilities. We are applying AI to help some of the largest Enterprise companies transform their businesses. We helping the largest Airlines better plan. Their schedules station and manage unforeseen. Operational disruptions. We helping Justice companies, achieve better features, libation route planning, and maximize the life of the assets. We helping Automotive companies, bring their release and three component, failure, and
recalls in real-time action, for adaptive control and visual inspection in more detail in the session. Your smoothie companies that are trusting Google to transform their industrial operations and businesses using Ai and machining Technologies knowledge. He's talked about where we are in the broader Industrial Automation evolution. Advances in automated controls and investment-wise yielding Lena return to point of platinum is not combined with a dynamic and an environment, like the one we finding ourselves right now in Market forces
precious to stay afloat. Industrial Systems and unadoptable their search for efficiency, gains with yours or systemize rules and leaders of operational buffers and inefficient Eve, and accrued by human expertise Industrial Systems. A great candidate for Solutions. Machine Learning Systems. Each layer of complexity is another dimension to explore for hidden correlations, allowing every aspect of an industrial environment to be analyzed, minutely him poorly and copy literally and Unbound, efficiency gains.
The system can adapt to internal and external headwinds and learn to anticipate allowing stable, autonomous operation. Will you talk to two such examples of autonomous control? Technology adapter controls and visual inspection, AI in industrial facilities. list of US talked about industrial adapter, controls platform, that automates and optimizes, closed-loop Industrial Systems, or Platform bills on years of Google's research in reinforcement, learning AI agent directly from sensory
experience and develops, new bikes blowing the actual space and then optimize is based on the most efficient future predicted state with no handcrafted rules a general-purpose learning framework and produced excellent recommendations. Several industrial optimization problems. Adidas, enter with actions and measurable rewards, especially the large pieces of equipment with billions of ways of operating them simply by system. With 10 pieces of equipment each which has 10 different set points.
That's already a million different combinations with each other in normal in your manners, while environmental conditions, such as I T load, and whether a constantly changing controls platform has been helping reduce energy consumption in Google data centers through Direct. So how does this work? As a reward function, this could be energy consumption or any other function expressed as a mathematical equation of sensors while adopting to be changing variables and still keep meeting operational constraints like building comfort.
For example, in the data center scenario evaluate millions of allowable setpoint combinations for future, states of energy, consumption, and operational. Constraints removes those that violate equipment or safety constraints chooses the set point that maximizes, the objective function Returns. The recommended set point in do a BMS platform to be implemented directly into the equipment to be sold using this approach. Imagine the impact that this would have on the global energy used in airports, corporate building weight
reduction in chemical plants metal processing facilities. Hire Industrial Another problem space where when you're playing Google's AI capabilities is industrial detected. Action is computer vision that use preaching machine learning model on image receptor in model. One step further with you partnering with some of these largest manufacturers of the world's to bring high-accuracy premium models for the backseat action inspection and cosmetic inspection. In an end-to-end automated
solution. We've created a dedicated breakout session to cover the technical breakthroughs of this technology. Delivered by one of my teammates Ying FEI, please do check it out. Ellie is a blight on machine learning models to achieve 99.9, accuracy for cosmetic defect reduction in the glass display panels. Globalfoundries have used our machine learning models for silicon Vapor detected action achieving a 95% accuracy across countries. Which makes for a good segue into our final
topic of stealing a global industrial scale and it's a completely different thing transforming in, in Tire Industrial Products used to be a privilege of deployment at a large-scale. We've come to embrace that the real-world environments are highly complex, unstructured and fragmented. This reality has made a design, a diploid. And even approach Legacy systems and devices from cloud hybrid Edge, from microcontrollers CPUs, and gpus CPUs, to clusters of servers
industrial equipment. Like I speed cameras and drones and Hardware supporting mqtt and OPC UA industrial standards. In working with the largest Enterprise, customers machine learning models in production, in one that allowed to be deployed on any kind of device. These Services allow you to work and control monitor performance and retrain models across a large Fleet of vices You can done more about these deployment Platforms in the dedicated breakout sessions. You can also connect with any of our machine learning Specialist or contact us through our website to
get started. Why do I feel excited that he is as a discipline? Has reached a stage where it's having a direct and measurable impact on stealing our Collective human outfit. Our Technologies are truly living up to the promise of furthering our progress as a society and fearing a more promising future. Thank you. Once again for joining and I hope you enjoy the other next session. Have a great day.
Buy this talk
Buy this video
With ConferenceCast.tv, you get access to our library of the world's best conference talks.