WELCOME [sercuarc.org]€¦ · 2020-06-03 · DevOps (2009) Cyber-Physical Systems (2006) Internet...
Transcript of WELCOME [sercuarc.org]€¦ · 2020-06-03 · DevOps (2009) Cyber-Physical Systems (2006) Internet...
SERC Talks June 3, 2020 1
WELCOME
Today’s session will be recorded.
An archive of today’s talk will be available at: www.sercuarc.org/serc-talks/ as well as on the SERC YouTube channel.
Use the Q&A box to queue questions, reserving the chat box for comments, and questions will be answered during the last 5-10 minutes of the session.
If you are connected via the dial-in information only, please email questions or comments to [email protected].
Any issues? Use the chat feature for any technical difficulties or other comments, or email [email protected].
How Do We Make Sure Future Cyber-Physical Systems Are Human Centered, and In Particular, Trustworthy?June 3, 2020 | 1:00 PM ETDr. Martin Törngren, Professor, Department of Machine Design (MMK), KTH Royal Institute of Technology
Towards Human-Centered and Trustworthy
Cyber-Physical Systems
3
Martin Törngren, [email protected]; www.kth.se/profile/martint
Mechatronics and Embedded Control Systems, Machine Design, KTH - Royal Institute of Technology
• Sustainable Industry and Society – the IRIS initiative• Centers
Trustworthy Edge Computing Systems and Applications
KTH & industry competence network - www.ices.kth.se
The lecturer - Dr. Who?
• Background in Mechatronics and Embedded control systems• Professor at KTH in Embedded Control Systems 2002
• Autonomous CPS - architecting, safety and MBSE • Sabbaticals at UC Berkeley and Stevens Institute of Tech.• CPS roadmapping
4Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren
Human-centered Cyber-physical systems?
Elaborated from FE Bureau: www.financialexpress.com/education-2/what-will-shape-future-smart-cities-of-india-find-out-here/968678/
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 5
Millions of tons of CO2 emissionsfrom 1850 with projection
Outline
Towards Human-Centered Trustworthy CPS
4. Directions3. Complexity analysis
1. Automotive and Automated Vehicles
2. CPS generalization andimplications
5. Conclusions
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 6
Mechanics Mechatronics
XXXDriver
XEng
XXClutch
XXTrans
XXDiff
XXXXWheel
XXSteer
XXBrake
XXSusp
DriverEngClutchTransDiffWheelSteerBrakeSusp
X
PX+PPPX+PX+PPDriver
PX+PPPPPPPEng
PX+PX+PPPPClutch
PPX+PX+PPPPPTrans
PPX+PX+PPPPDiff
XX+PXXWheel
X+PPPPPX+PPPSteer
X+PPPPPX+PPPBrake
X+PPPPPX+PPPSusp
DriverEngClutchTransDiffWheelSteerBrakeSusp
Design structure matrices illustrating vehicle components and relations: P – Programmable relations. X – Possible change
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 7
10100011010011100101001110100011010011100101001110100011010011100
101001110100011010011100010011
Organizational complexity
Production/parameterization
ProjectManagement
Sensing,estimation
and controlIntegration tests
CAD, PDM
ECU/electronics
Quality
ArchitectureNetworking
CAN database
Diagnostics,Support tools
Safety
Maintenance,After-market
Vehicledynamics
Software
• Power• Propulsion• Thermal
Mechanical components
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 8
Transportation melting pot
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 9
Digital infrastructure and connectivity
Telecommunication: …3G, 4G, 5G, …Smart phones/padsWireless and wired communicationInternet and cloudIndustrial computingSmart devices and embedded systems
The world as a connected and
distributed system
Courtesy of Ericsson
www.energimyndigheten.se/tester/tester-a-o/elmatare-smarta
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 10
Edge Computing systems – with severalinterpretations
Cloud Computing
Embedded nodes Internet / MobileNetwork
Edge nodes
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 11
Drivers: - Real-time ”AI” augmentation (e.g. prediction, coordination)- Cost-efficiency, bandwidth and energy- Privacy
Autonomy: Basic functions
Observation LocalizationObject detection& classification
Planning and generation
Control
Steering Propulsion Braking
Energy managementVehicle dynamics
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 12
Unprecedented capabilities and complexity
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 13
Breaking new ground – generalized knowledge
ADI – ”Autonomous Driving Intelligence”
Illustration: Harry Campbell, IEEE Spectrumhttp://spectrum.ieee.org/cars-that-think/transportation/self-driving/nxps-bluebox-bids-to-be-the-brains-of-your-car
By Veronica538 (Own work) [CC BY-SA 3.0 (http://creativecommons.org/licenses/by-sa/3.0) or GFDL (http://www.gnu.org/copyleft/fdl.html)], via Wikimedia Commons
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 14
What drives AV development?
Business MIT Tech Review:- Can We Put a Price on Autonomous Driving?
• Transport services: ~ Trillions of dollars!• Traffic accidents: 100’s of billions of dollars• Traffic efficiency, productivity and public health: - II -
The beginning of wisdom is to call things by their proper name– attributed to Confucius
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 15
1849 – Gold rush
Today: Billions poured in toget to the Trillions!
Key question:When will the Gold emerge?
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 16
Socio-technical impact
Automatedvehicle(s)
:
s
Societal aspects, ethics, values
Technology, methodology, standards
Research & education
Transportation systems
Desired propertiesLife-cycle operation
Legal frameworks
Business eco-systems &organizations
Human machine
Software engineeringElectronics and mechanical engineering
Robotics, Artificial intelligence & machine learningEmbedded and high performance computing
Safety, security and dependabilityPower and energy systems
Systems engineeringTowards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 17
Outline
Towards Human-Centered Trustworthy CPS
4. Directions3. Complexity analysis
1. Automotive and AVs 2. CPS generalization andimplications
5. Conclusions
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 19
Digitalization in a physical world- with concepts mirroring different perspectives
Edge/Fog (~2012)DevOps (2009)
Cyber-Physical Systems (2006) Internet of things (1999)
Ubiquitous computing (1988)Artificial Intelligence (1956)
Cybernetics (1948)Teleautomation (1926)
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 20
CPS etymology
cyber - the use of computers or computer networks [Merriam-Webster] – with many connotations; cyber-space, -security, ...
cyber – “coined” by Norbert Wiener in the 1940’s, from the Greek word “kubernetes” - "governance", referring to feedback systems!
Both interpretations make sense for CPS! • interactions and integration among “C” and “P” elements!• Involving sensing and/or control for “C” and “P” interactions• Humans as an additionally important element!
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 21
CPS characterization (from the CyPhERS project)
• Physical vs. Embedded/Edge/Cloud vs. Communication• Scale, Openness, Mobility• Capabilities, level of automation, hierarchy, decentralization• Role of humans - In-/Outside the Loop• Degree of life-cycle connectivity• Governance and Jurisdiction
• Single Domain vs. Cross Domain • CPSoS – multi-organizational
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 22
Cengarle M.V. and Törngren M., et al. Structuring of CPS Domain: Characteristics, trends, … associated with CPS. Deliverable D2.2, CyPhERS project, May 2014
Cyber-physical systems (~2006)
Integration of computation, networking and physicalprocesses where CPS range from minuscule (pace makers) to large-scale (e.g. national power-grid).
Not new but with• Improved and entirely new capabilities• Increasing integration across technologies, systems,
domains and life-cycle• Open society scale deployment• Business model evolution
Unprecedented opportunities and risks
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 23
CPS capabilities
Gather, store and process all kinds of dataAwareness and prediction, including “mind reading”Plan and make decisionsAffect and create physical as well as software systemsGenerate and control energyCollaborate - exchange information, visualization, AR/VRCapabilities as part of “loops” with various time and system perspectives (e.g. OODA, MAPE-K)
Reinforced by multiple progressing technological fronts Unprecedented opportunities and risks!
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 24
Snapshot of benefits and concerns/risks
How do we consume as much ofyour time and attention as possible?… dopamine hits (Sean Parker)
Air bags - 1968ABS - 1970
ESP - 1982AEB – 2012
”… the mere prescence of one’sown smartphone reducesavailable cognitive capacity”
Threats and risks
25
Deep fakes, inspection, Digital surveillance economies
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 25
Safety/Assurance cases
Safety case: “… a structured argument, supported by a body of evidence, that provides a compelling, comprehensible and valid case that a system is safe …” - NASA System Safety Handbook ver. 1 (2014)
Safety cases are non trivial even for current systems• Boeing 737 MCAS; Current level 3 systems in testing • Automation paradox will not go away!• Socio-technical!• Safety requires a constant struggle!
2626
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 26
Herbert Simon on industrial revolutions
• There is no single technology behind a revolution!• Technological revolutions are slow
• But the disruption can be hard to anticipate!• Digital platforms enable faster market launch• The first revolution (involving steam engines) took 150 years
to change society (6 generations)• We shape technological revolutions!!
– Social impact becomes tangible in the 2nd half
Amara’s law: We tend to overestimate impact in the short run, but underestimate in the long run
H. Simon: The steam engine and the computer
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 27
Several parallel gold rushes!
Strong market forces!Partly unregulated areas!
1200? IoT platforms
Automated driving/transportation services
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 28
Disruption and change
Fortune 500 top list changing more frequently• Since 2000, more than 50% of the companies have
disappeared due to the digitalization wave
http://www.internethistorypodcast.com/2016/07/inventor-of-the-first-digital-camera-steven-sasson/
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 29
Largest internet companies wikipedia, Jan 2020)
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 30
CPS – as representing a socio-technicalrevolution
Learning from previous technical revolutions- Time constants, multiple technologies, ”shaping”- Connected collaborating smarter electrified CPS CPSoS, - Data and service age- Strong economical drivers and disruption- Evolving technologies and business eco-systems- Changing behaviors, systemic and socio-technical effects- Complexity reflecting new capabilities and open environments
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 31
Outline
Towards Human-Centered Trustworthy CPS
4. Directions3. Complexity analysis
1. Automotive and AVs 2. CPS generalization andimplications
5. Conclusions
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 32
How can we handle the increasingcomplexity?
Source: https://en.wikipedia.org/wiki/Human_brain
Visualization of software and electronic functionalities in a truck (2013) Courtesy of Scania
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 33
Source: www.techiestuffs.com/trend-of-computer-aided-engineering-worldwide/
CAE support
The degree of difficulty in accurately predicting behavior over time
CPS complexity
facets
ORG. limitations
Consequences of complexity(on humans and projects):Difficulty to understandDifficulty to predict and trade-off EmergenceProject overruns (cost and time)Requirements not metResources & competences needed
Facets causing humans toperceive complexity
Environment
characteristics characteristics
CPS ORGs
Heterogeneity/diversitySize and computabilityUncertainty and changeDynamics and/or StructureIncidental/essential
Human memory capacityBounded rationality&biasesHumans: autonomous agentsCommunication barriersInformation capture OHInformation managementInteroperability barriers
Means:- Process &
organizational-Design &
architecting- MBE /CAE- Business models
- Software and data- Interrelations management- Competence and Life-long learning
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 34
M. Törngren and P. Grogan: How to Deal with the Complexity of Future Cyber-Physical Systems? Designs, 4, 2018
Physical vs. Software vs. Data systems – a broader view is essential!
Physical systems Software DataPhenomenaand effects
Multiple coupled physicalphenomena, (wear, fatigue, heat, …)”slow cycles”
State space; bugs; connectivity; variabilityFast local & global effects”very fast cycles”
Super-human performance; Brittleness
Abstractions, synthesis, and platforms
Approximations, Cont. time & value; No single “platform”, Behavioral model sim. vs. Geometry based synthesis
Digital / discretization”platform” foundations; Logic preserving transf.; Abstracted physical props!
New programming model; model- or learning based(“ML”, “RL”, “DL”)
Extra-functionalproperties
Established cost models Dependencies and reusecreate additional relations.Difficult to estimate life-cycle cost
SW under the hoodFailure modes; Quality, availability; accessibility; quality
Combined … and in systems at different levels
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 35
Human cognitive abilities and limitations
Short-term MEM
Long-term MEM
Actions/proceduresConceptsAssociations
Sensors
Cere-bellum
”slow system”
Narrating selfHeuristics and biases
Capacity: ~ 5-10 ”chunks”
Limited budget ofattention & interference
Learning and adaptation
Potentialoverruling
Super-vision
”Cache”
Spinal cord
Actuators
”fast system”
We tend to divide separate aspects into categories (Sapolsky)• Boxing helps focus but is devastating to ”thinking out of the box”• Viewpoints are arbitrary; Group think
Difficulty in dealing with feedback and delayed systemsWe are not primarily rational beings (Kahneman)
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 36
Outline
Towards Human-Centered Trustworthy CPS
4. Directions3. Complexity analysis
1. Automotive and AVs 2. CPS generalization andimplications
5. Conclusions
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 37
Industrial and societal transformation -towards sustainability
Leverage CPS and digitalization for Sustainability• Transport, agriculture, energy …
New methodologies to engineer CPS, towards• managed complexity, ensuring trustworthiness: H-CPS• mitigate side-effects, unintended emergence and unreasonable risk
What can various actors do?• Collaboration, open data sharing, testbeds, • Policies and incentives: balancing innovation & risk• Education
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 38
CPS example – opportunities and co-design
Fuel-cellDistributed controlElectrical actuators
Hy-Wire Skateboard conceptIn 2002 (GM)
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 39
CPS capabilities and design towards sustainability
Leveraging CPS and CPSoS design: • Replacing/reducing CO2 emissions
– “system level effects”• Energy use and management
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 40
Gather, store and process dataAwareness and predictionPlan and make decisionsAffect and create physical & SW systemsGenerate and control energyCollaborate - exchange information, visualizationAt various time and system perspectives
Towards a circular economy• Tracing, identification, prediction• Modularity• Individualized tailored production of spare parts• Reuse, recycling, upgrading, downgrading, maintenance
Modeling and simulation tools
Co-simulation FMI, HLA, DDS, ModelCONNECT, AD-EYE
KTH survey on Modeling & sim. tools:Dated: ”Jan. 2019”!Courtesy: Xinhai Zhang
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 41
A need for new CPS methodologies- to manage complexity and incorporate sustainability
Design for X – integrated design – Co-design• Humans - CPS; Trustworthiness properties; Cyber & Physical • Energy, Materials, Digitalization, Resilience, Learning, Incentives
Supporting humans in dealing with complexity • Dealing with large design space and complex environments! • Embedded situation awareness and risk metrics/management• Visualization - AI-Assistants• Collaboration and communication skills• Awareness: Complexity; knowledge on human cognitive abilities
The Cynefin model
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 42
Anchoring and bridging viewpoints
43
Con
cept
Leve
lM
odel
Leve
lTo
ol/
cmpo
nent
ConceptLevel
Model
LevelTool/
cmponent
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 43
M. Törngren et al. Integrating Viewpoints in the Development of Mechatronic Products. Mechatronics, Dec. 2013
Resilience principles from social-ecological systems
Maintain diversity and redundancyManage connectivityManage slow variables and feedbacks Foster complex adaptive systems thinkingEncourage learningBroaden participation Promote polycentric governance systems
Book: Principles for Building Resilience -Sustaining Ecosystem Services in Social-Ecological Systems, 2015
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 44
Education and training challenges
Need for educational renewal!
Need for life-long learning!
Low status of education is a problem!
Role of competence networks!
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 45
Martin Törngren et al. Competence networks in the era of CPS – lessons learnt in the ICES cross-disciplinary and multi-domain center. Workshop on Embedded and Cyber-Physical Systems Education (WESE), 2019
How is KTH responding to these challenges?
• Emphasizing educational renewal and life-long learning• Broader view of competence
• Testbeds for future CPSoS, including• Intelligent transportation
• Smart homes – Live In Lab: www.liveinlab.kth.se/en
• Competence networks: students; faculty; industry• Sustainable Industry and Society – the IRIS initiatve• Digitalization and Trustworthy CPS
• Trustworthy Edge Computing Systems and Applications
46Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren
Outline
Towards Human-Centered Trustworthy CPS
4. Directions3. Complexity analysis
1. Automotive and AVs 2. CPS generalization andimplications
5. Conclusions
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 47
How can we trust future CPS?
Failure modes, explainability, understandability, ethics/valuesSafety, security, reliability and privacy – multiple attributes!
The Automation paradox!
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 48
Towards human-centered and sustainablecyber-physical systems!
Elaborated from FE Bureau: www.financialexpress.com/education-2/what-will-shape-future-smart-cities-of-india-find-out-here/968678/
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 49
Millions of tons of CO2 emissionsfrom 1850 with projection
Selected references
Martin Törngren et al. Competence networks in the era of CPS – lessons learnt in the ICES cross-disciplinary and multi-domain center. Workshop on Embedded and Cyber-Physical Systems Education (WESE) at ESWEEK 2019, SpringerMartin Törngren and Paul T. Grogan. How to Deal with the Complexity of Future Cyber-Physical Systems?, Journal of Designs, Vol. 2, No. 4, 2018 Martin Törngren and Ulf Sellgren. Complexity Challenges in Development of Cyber-Physical Systems. In Principles of Modeling; M. Lohstroh et al – editors; Springer, 2018; Vol. 10760, Lecture Notes in Computer Science, July 2018Martin Törngren et al. Architecting Safety Supervisors for High Levels of Automated Driving. 21st IEEE Int. Conf. on Intelligent Transp. Systems, Nov. 2018.Martin Törngren et al. Integrating Viewpoints in the Development of Mechatronic Products. Mechatronics, special issue on Model-based mechatronic system-design, Elsevier Dec. 2013Patricia Derler, Edward Lee, Martin Törngren, Stavros Tripakis. Cyber-Physical System Design Contracts, ICCPS '13: ACM/IEEE 4th International Conference on Cyber-Physical Systems, 2013Platform4CPS recommendations: www.platforms4cps.eu/fileadmin/user_upload/E-Book_-_Platforms4CPS_Key_Outcomes_and_Recommendations.pdf
Towards Human-Centered and Trustworthy Cyber-Physical Systems, June 2020, Martin Törngren 50
Herbert Simon. The Steam engine and the computer. EDUCOM Bulletin Vol. 22, no. 1, 1987)Daniel Kahneman. Thinking, Fast and Slow. Macmillan, 2011Robert M. Sapolsky. Behave - The Biology of Humans at Our Best and Worst, 2017
SERC Talks June 3, 2020 52
SERC Research Review 2020: Save-the-Date
For more information, contact Monica Brito: [email protected]
Visit our website for more information: https://sercuarc.org/research-reviews/
SERC Talks June 3, 2020 53
UPCOMING TALKS:“Mission Engineering” Series
Tentative Dates:Wednesday, August 5, 2020Wednesday, October 7, 2020
Wednesday, December 2, 2020
CONTACTEditor-in-Chief: Dr. Barry Boehm, University of Southern California – [email protected]
Webinar Coordinator: Ms. Mimi Marcus, Stevens Institute of Technology – [email protected]
Please visit the SERC Talks page to register and for more information and updates.
SERC Talks June 3, 2020 54
Thank you for joining us!Please check back on the SERC website for today’s recording
and future SERC Talks information.
Subscribe and follow SERC on our social channels: