Omron AI Controller Learning Session Updated 1182018 Confidential

Omron AI Controller: Learning Session Updated: 11/8/2018 Confidential B

Why am I attending this? • Know what it does, and the talking points • Know why your customer is a target • Know your next steps in evaluating opportunities

What is the Omron AI Controller? New solution that is uniquely able to Collect Analyze Utilize data on the Edge within a Sysmac controller, for the purpose of extending equipment lifecycle NY-Series NX-Series AI Application Component Configure & Startup

Cloud Fog What is an “Edge” device? Within the Machine or Production Line, without any separate Windows® PC or remote server (Cloud) or Internet Computing Edge Cloud Fog Edge Collect Analyze Data Collection Processing Storing Modeling ✔✔ ✔+ ✔+ ✔ ✔- Utilize Real-time Monitor Control FB Long-term Monitor Visualize ✔+ Real-time Monitor Control FB ✔ ✔+ Long-term Monitor Visualize ✔- Real-time Monitor Control FB

What does the Omron AI Controller do? Detecting outliers (failures, defectives) by learning from historical data in the machine 1) Issue definition 2) Cause identification Machine defects 3) Prepare sensors Method Human B occurs Material Based on failure impact and feasibility study of AI Controller utilization, define issue Machine Based on Cause and Effect Diagrams, identify cause of machine defect Based on the cause, add necessary sensors to machine Human Omron AI Controller fully covers the data collection, analysis, and utilization. Omron also offers startup support for the entire process. 4) Collect Data collection Pre-processing 6) Utilize 5) Analyze Storage Causal analysis, modeling X 1 Monitoring Visualization Control FB X 2 X 5 Machine X 3 Collect the raw data of machine Eliminate noise, create and collect the features of normal/abnormal state Store the characteristics data X 4 Create model data after causal analysis Status monitoring and control FB based on model data

Machine Learning Anomaly Detection Learn without being explicitly programmed

First Rele ase AI Application Components Pre-made Sysmac Library Function Blocks (FB) for… Mechanism Air Cylinder Ball Screw Conveyor Belt or Pulley Failure Type Machine event (failure causes) Operation stop Rubber Gasket broken (Rod/Piston) Over speed Speed Controller broken Over vibration Air Cushion broken Operation stop Contamination Operation stop Guide broken Operation stop Ball Bearings falling out Operation stop Contamination Vibration and low accuracy Contamination Operation stop Belt loosen Operation stop Belt broken Operation stop Pulley broken

So what? 1) Detect an issue has happened and do something about it immediately, reducing risk of bad parts or equipment damage 2) Leverage Omron’s technology and Data Scientist team instead of the customer figuring it out on their own 3) No extra infrastructure/connectivity cost for Fog or Cloud based solutions (typically managed by IT Department)

OEE Overall Equipment Effectiveness Value of Omron AI Controller Helps improve OEE at the machine level by increasing uptime = Uptime Operating time Stoppage Loss Availability Downtime x Performance Quality Net operating time Defect Loss Value operating time Performance Loss x Loss types Improved by AI Controller 1: Equipment failure 2: Changeover 3: Part replacement 4: Waiting for material 1: ✓ 2: 3: ✓ 4: 5: Minor stoppage/idle 6: Deceleration/slowing 5: ✓ 6: 7: Defect/modification 7: ✓

Why my customer? Why now? • Strategic Account who has already shown interest • Regional Account who has already purchased baseline hardware • Pre-launch assessment before December* Launch

2 H FY 18 Target Accounts q q q q ABBOTT VASCULAR q q AMERICAN ORTHODONTICS CORP q AUTOLIV ELECTRONICS CANADA INC q AUTOMATION SOLUTIONS INC. q BIMBO q CORNING INC. q CSU LEON YOUNG WAREHOUSE q DMAX LTD / SETECH SCS LLC q FALCON HI TEC INTL INC. q FANUC AMERICAS CORPORATION q FOXCONN ASSEMBLY LLC q HEGLAR CREEK ELECTRIC LLC HITACHI AUTOMOTIVE SYSTEMS MEXICO SA DE CVq q HONDA q INDUSTRIAL CONTROL AUTOMATION q ISTHMUS ENGINEERING JTEKT KD AUTOMATION INC LABPLAS MAIBEC INC. NEMAK MEXICO S. A NESTLE NEW CONCEPT TECHNOLOGY NISSAN DECHERD NSK CORPORATION OLIN WINCHESTER CENTERFIRE PHILIPPE MERCIER INC. SPECTRUM BRANDS INC T S TECH USA CORPORATION TOYODA GOSEI XIGENT AUTOMATION SYSTEMS INC ZIMA-PACK LLC Confidential

Types of Customers & Applications Customer Types who can benefit from this technology 1) 2) Highly specialized product with expensive or complex process that would benefit from Anomaly Detection to maintain high product quality or reduce costly scrap/line-damage, typically not interested in this data going to alternative Cloud based services due to high confidentiality of IP regardless of speed of utilization Ø Example: Abbott, ANBA, Foxconn, Honda, Nissan, etc. Very high volume production of relatively inexpensive or simple product that would benefit from Anomaly Detection to reduce even the slightest production stoppage or save some small cost per each product, typically not able to use alternative Fog/Cloud based services due to high utilization speed requirements Ø Example: Bimbo, Isthmus, Nestle, Zima-Pack, etc. v. Must have some acceptance of statistical correlation and machine learning as a concept in order to accept technology, comfortable with setting up threshold-based automated decisions, and clear problem statement that can be understood by different levels in the organization to minimize mistrust/sabotage Ø Example: low number of false negatives is critical to gain acceptance by production managers & operators or else they will find way to bypass/ignore system Confidential

Cost & Pricing • Hardware premium is not enough to be sustainable • Start-Up Service and periodic Paid Service • Annual Subscription on customer’s ability to adjust their own Thresholds, and access to new AI Application Components All Fog & Cloud based solutions are Annuity based business models, but customer feels Trapped

Next Steps… 1) “Understand” material from Learning Session 2) Does the target customer have an upcoming opportunity using Sysmac, or would they be willing to upgrade their existing Sysmac NX 7 or NY 5? q Yes Proceed to Step 3 q No Proceed to Step 5 3) Could the opportunity benefit from Anomaly Detection on the target mechanisms (Air Cylinder, Ball Screw, Belt/Pulley)? q Yes/Maybe Proceed to Step 4 q No Proceed to Step 5 4) Rejoice – contact Mike Chen for additional steps 5) Maybe not the right timing now, still close-out with Mike Chen

What’s In It For Me? • Larger more unique solution sell bringing totally new value • Capture customer’s year-end spend with service annuity • Glory of selling the first-of-its-kind in the industry

Gotcha’s • • “Predictive Maintenance” – the technology enables humans to make predictions “Continuous” Learning – No, it does not do that (yet) Anomaly Detection – Yes, that is all it does (so far) Existing Sysmac user – fastest sell, focus only on new capabilities & value Data Concentrator – theoretically possible, but needs further clarification Custom AI Models – longer sales process, but still want to get feedback Annuity Business – looking for recurring paid service opportunities Promotional Material – still in progress until official launch Confidential

Thank you Questions & Feedback – contact Mike. Chen@omron. com