Accelera�ng AI in Edge Compu�ng - Accelerate your path to industrial Edge AI applica�on - Improve manufacturing by AI with Edge compu�ng Smart Factory Solu�on www.mitacmct.com / sales_client@ mic.com.tw / +886 3 3

How to Accelerate Your Path to Industrial Edge AI applica�on? The data analy�cs driven by AI are transforming industrial segment and different insights are generated from basic, opera�onal analy�cs to more advanced analy�cs. Here is the path to deeper AI analy�cs: 3 Predic�ve Analy�cs Foresight (What will happen, when and why?) 2 Diagnos�c Analy�cs Insight (What happened and why?) Prescrip�ve Analy�cs Forecast (How should I proceed?) Cogni�ve Analy�cs Self-learning (How do I proceed?) Connect things and collect data; visualiza�on of assets Anomaly detec�on, defect detec�on, compliance monitoring,...

Right Hardware Pla�orm pla�orm (like Intel) has a por�olio of products for AI compu�ng, from workload-focused CPU and GPU products to programmable components and deep learning accelerators. Intel's CPUs are the most flexible processors for most generic AI workloads, and are the founda�on for deep learning inferencing with AI enhanced capabili�es (e.g. Intel® Deep Learning Boost, Intel® AVX 512) integrated in the silicon. Intel's integrated GPUs with hardware support for AI workloads built in (e.g. dp4a) enhance the overall CPU goodness for AI workloads. Intel® Movidius™ VPUs enable computer vision...

In order to know about the CPU, GPU and different GPU difference, the sec�on below will discover more about it. By knowing this fundamental mechanism, we are more available to choose the right hardware configura�on for each of our aiming Edge AI applica�ons. What Is a GPU? Why GPU for AI? Modern GPU technology powers tradi�onal graphics applica�ons—and much more. GPUs were originally designed to accelerate the rendering of 3D graphics. Over �me, they became more flexible and programmable, enhancing their capabili�es. This allowed graphics programmers to create more interes�ng visual effects and realis�c...

How to Build So�ware Applica�on Exis�ng proprietary middleware solu�ons that are inflexible and vendor-locked limit the op�ons for industrial solu�on providers. These solu�ons impose a high total cost of ownership and require �me and resource for so�ware stack development, and therefore make scaling challenging. The challenges could be like: Market gaps with exis�ng solu�ons Locked and proprietary, non-flexible High total cost of ownership Scalability performance requirements Data privacy Lack of general purpose pla�orm for mul�ple analy�cs use cases Solu�on providers' painpoints to build Lack...

Therefore, here is how you can advancing your compe��ve differen�a�on whit Intel’s Edge Insights for Industrial So�ware unique features: Cloud Agnos�c Pick and choose desired micro-services for specific customiza�on needs Run analy�cs via Intel developed or 3rd party algorithms Connect to any CSP to take advantage of complementary cloud-edge solu�ons MiTAC is Intel selected partner to enable EII capability in MiTAC team and also pre-installed it in several of our embedded hardware systems as a demo kit for worldwide domain SI vendors to use and to try. All of these is aiming to lead the industry...

Manufacturing AI with Edge Compu�ng Currently data collected is used mostly for the detec�on and control of any abnormali�es. However, the goal going forward is for the op�miza�on and predic�on of issues to be faced in the whole manufacturing process. Edge compu�ng allows synchronous opera�ons for these typical cri�cal processes such as data analysis, decision making and remedial ac�ons. Likewise, developments in cloud technology have cut costs and helped users focus on their core processes. Virtualiza�on provides the agility required to speed up IT opera�ons and reduces cost, by increasing infrastructure...

Use Cases of Edge Compu�ng in Manufacturing Machine Vision - Early interven�on - Increase produc�vity - Reducing in the cost of low-level labor - Improve accuracy and reduc�on in scrappage - Reduc�on in the space required for the plant (manpower opera�on area to machinery and equipment area) The sensi�vity and the resolu�on in vision systems have the ability to detect weak impulses at invisible wavelengths in dim light and differen�ate between objects. It can help determine where the problems are being introduced in a produc�on line so correc�ve ac�on can be taken. Ar�ficial intelligence techniques...

Use Cases of Edge Compu�ng in Manufacturing - Prevent product damage - Reduce labor costs - Increase produc�vity and automate processes - Used for picking, transporta�on, and the ability to sort without any manual interven�on. To automa�ze a factory or warehouse, AGV will help be required where there is a need for repe��ve, consistent material deliveries, and where large ini�al cost outlays and long return on investment (ROI) can be tolerated. However, AGVs are being challenged by evermore sophis�cated opera�ons and need for flexibility. MP1-11TGS with the powerful CPU, but low power consump�on...

Use Cases of Edge Compu�ng in Manufacturing Predic�ve Maintenance - Reduced MTTR (Mean Time To Recovery) - Fewer machine failures - Reduce maintenance costs - Increased asset life�me - Improve workplace safety Being a step ahead, before a problem arises, is the goal of predic�ve maintenance and risk preven�on. Predic�ve maintenance is to localize poten�al equipment malfunc�ons, to avoid more complex maintenance ac�vi�es, by data collec�on to find a pa�erns and analyse the tailored maintenance rou�nes. To optimize the prediction, these machines need to be intelligent enough to recognize that they...

Use Cases of Edge Compu�ng in Manufacturing Produc�on Op�miza�on - Predict produc�on performance varia�ons - Increase factory transforma�on with AI - Reduce down�me or over�me and any unscheduled maintenance - Analyze sources of waste with AI - Boost quality and yield performance Factory automa�on increases manufacturing efficiency and the reduce costs, accelerates the revolu�on in produc�on processes. Typical problems like variable raw material quality, intricate plant layouts, monitoring of material in real �me, inconsistent opera�ve procedures, dis�nct designs across loca�ons, constrained adaptability...