Presented at the Panel on
Sensor, Data, Analytics and Integration in Advanced Manufacturing, at the Connected Manufacturing track of Bosch-USA organized "Leveraging Public-Private Partnerships for Regional Growth Summit". Panel statement: Sensors, data and analytics are the core of any smart manufacturing system. What are the main challenges to create actionable outputs, replicate systems and scale efficiency gains across industries?
Moderator: Thomas Stiedl, Bosch
Panelists:
1. Amit Sheth, Wright State University
2. Howie Choset, Carnegie Melon University
3. Nagi Gebraeel, Georgia Institute of Technology
4. Brian Anthony, Massachusetts Institute of Technology
5. Yarom Polosky, Oak Ridget National Laboratory
For in-depth look:
Smart IoT: IoT as a human agent, human extension, and human complement
http://amitsheth.blogspot.com/2015/03/smart-iot-iot-as-human-agent-human.html
Semantic Gateway: http://knoesis.org/library/resource.php?id=2154
SSN Ontology: http://knoesis.org/library/resource.php?id=1659
Applications of Multimodal Physical (IoT), Cyber and Social Data for Reliable and Actionable Insights: http://knoesis.org/library/resource.php?id=2018
Smart Data: Transforming Big Data into Smart Data...: http://wiki.knoesis.org/index.php/Smart_Data
Historic use of the term Smart Data (2004): http://www.scribd.com/doc/186588820
Market Analysis in the 5 Largest Economic Countries in Southeast Asia.pdf
Smart IoT for Connected Manufacturing
1. Smart IoT for Connected Manufacturing
for Bosch Summit: Connected Manufacturing panel on
Sensor, Data, Analytics, and Integration in Advanced Manufacturing
September 17, 2015; CMU-Pittsburgh, PA
Amit Sheth
Ohio Center of Excellence in Knowledge-enabled Computing (Kno.e.sis)
Wright State University, USA
Special Thanks: Pramod Anantharam, Tanvi Banerjee
2. 2
1The Always-On Assembly Line: How Fault Tolerant Servers Boost Output and Reduce Costs for Manufacturing
2Reliability magazine, 2002.
3White paper: Reducing Operations & Maintenance Costs, 2003
$44,000
$60
billion
$1.5
million
Lost per minute when
an auto plant stops due
to failure1.
Spent on warranty
remediation in 2006 by
Auto manufacturers1.
Lost revenue per hour
downtime for U.S.
manufacturers in 20001.
Manufacturing Failures: Severity of the problem
86%
Maintenance is reactive (too
late) or preventive
(unnecessary) resulting in sub-
optimal savings2.
“Many of these problems could be reduced by adjusting the mix of reactive,
preventive, predictive, and proactive maintenance strategies so workers can
focus on doing the right things at the right time.3”
Flood of data and alarms during a fault overwhelms operators to
find and fix the problem in a timely manner3
3. Connected Manufacturing: Smart IoT as a Possible Solution
Smart IoT
Device and
Protocol
Interoperability
Data
Interoperability
Semantic
Interpretation
“… Smart Data a term initially proposed in
2004 but which is increasingly making sense in
conveying how all the volume, variety, velocity,
and veracity challenges of physical, cyber,
social Big Data needs to be managed to
derive value out of them.”4
Smart Data + IoT => Smart IoT4
4http://amitsheth.blogspot.com/2015/03/smart-iot-iot-as-human-agent-human.html
Ram D. Sriram, Amit Sheth, "Internet of Things Perspectives", IT Professional, vol.17, no. 3, pp. 60-63, May-June 2015, doi:10.1109/MITP.2015.43
4. Device and Protocol Interoperability
Heterogeneity in machines, devices,
sensors, and communication protocols
Semantic Gateway as a Service for
interoperability between devices that
are using different protocols
Pratikkumar Desai, Amit Sheth, Pramod Anantharam, 'Semantic Gateway as a Service architecture for IoT Interoperability', IEEE 4th International
Conference on Mobile Services, June 27 - July 2, 2015, New York, USA.
5. Data Interoperability: Annotation for Integrating Heterogeneous Data
Structured Data
• Device Status Data
• Alarm Data
• System Logs
Unstructured Data
• Work Orders
• Maintenance Notes
• Warranty Claims
Heterogeneity in data sources
encompassing both textual data and
sensor data
• Semantic Sensor Network
(SSN)Ontology for modeling
sensors and their observations
• Domain specific ontology such as
manufacturing tools, process, and
possible faults
Raw Data
Annotated Data
6. Semantic Interpretation: Processing Data for Actionable Information
How do we synthesize actionable
information from annotated data?
• Analogy5:
• Human sensory system sends 11
million bits of information per second
• Brain processes 16-50 bits per
second
• Semantic, cognitive, and perceptual
computing for deriving actionable
information
5source: Leonard Mladinow Subliminal
RDF OWL
?
7. Did you cough more
than 20 times today?
- Yes
Semantic Sensor
Network Ontology
Did you cough more than 20
times today?
- Yes
Did you cough more than 20
times today?
- Yes
Did you cough more than 20
times today?
- Yes
Did you cough more than 20
times today?
- Yes
Did you cough more than 20
times today?
- No
Raw Data
Historical Data from a Person
20 times cough is http://knoesis.org/asthma/high-coughing
672 steps is http://knoesis.org/asthma/low-activity
1 h 17 mins REM Sleep is http://knoesis.org/asthma/disturbed-
sleep
Annotated Data 672 is http://knoesis.org/asthma#steps
1 h 17 mins is http://knoesis.org/asthma#REM_Sleep
20 is http://knoesis.org/asthma#Cough_Incident
Personalizati
on
Well Controlled
Very Poorly
Controlled
Not Well
Controlled
Contextualization
Abstractions
(Actionable Information)
Knowledge Base
and Unstructured Data
The symptoms (high-coughing, low-activity, disturbed-
sleep) are interpreted with respect to a person with
severity level “Mild Asthma”
Interpretation Exploration Understanding
CC
PC
SC
Example from kHealth Project
8. Thank you
Thank you, and please visit us at http://knoesis.org
For more information on kHealth, please visit us at http://knoesis.org/projects/khealth
Cognitive
Computing
Semantic
Computing
Perceptual
Computing
Editor's Notes
Time series observations are readily and naturally available in domains such as finance, health care, smart cities, and system health monitoring. Increasingly, time series observations include both sensor and textual data generated in the same spatio-temporal context creating both challenges for dealing with heterogeneous data and opportunities for obtaining comprehensive situational awareness. For example, in a city, there are machine sensors and citizen sensors observing the city infrastructure (e.g., bridges, power grids) and city dynamics (e.g., traffic flow, power consumption). In this research, we investigate extraction of city events from textual observations and utilize them explain variations in the sensor observations. This will improve our understanding of city events and their manifestations due to the complementary nature of observations provided by the machine sensors and citizen sensors.
Device Interoperability
- SGAS
Data Interoperability
- Annotation using SSN and domain ontology
Contextual and Personalized Interpretation
- Semantic Perception