Artificial Intelligence #51: An opensource digital twin builder?

Welcome to #Artificial Intelligence #51

Our courses are getting filled up fast and Digital twins starts soon. If you want to study with us at #universityofoxford for our next course see:?Developing Artificial Intelligence Applications using Python and TensorFlow. Also, our course on?Digital Twins?at #universityofoxford is almost full.

This week,?I am covering a topic I have been working on for some time. As part of my teaching, especially the digital twins course as above, I have been formulating a broad methodology for AI in Cyber Physical systems. The subject is not easy to develop since it’s a complex / cross disciplinary topic

I have two motivations

Firstly, I wanted to create a general framework for teaching this subject especially for Digital twins (but now also adopted by our metaverse course and AI in Agtech course). Lars and I already use a motivating case study for the digital twins course. So, this framework serves as a palette to interconnecting complex concepts

?Secondly, I wanted to use it as a tool for developing AI and Cyber Physical systems.

?I prefer to call it ‘Cyber physical systems’ rather than industry 4.0 because we are covering more than manufacturing (ex we want to apply the same framework for agtech, climate change, urban mobility etc). Hence, a complex vision

?Cyber-Physical systems?encompass physical objects and infrastructure in a control mechanism through the Internet.?Artificial Intelligence (AI) for Cyber-Physical systems?amalgamates Edge, Autonomous systems, systems thinking, and design thinking.

Building on our roots of academia and research, our framework provides a palette to prototype and build industry use cases

?Technologies and concepts underpinning the digital twin are not new. In a nutshell, for a digital twin, you need the following:

a)????A digital entity that mirrors a physical object, process, or subsystem

b)????The ability to reflect the changes in the physical entity into the digital entity

c)????The ability to simulate the process in the digital domain and?learn from the simulation

There are some more developments

a)????The Digital thread – a twin which records the product or system across its lifetime

b)????AR/VR – as a simulation mechanism and a feedback loop

c)????Graph networks and Digital twins

d)?????Causality and Digital twins

e)????Physics based models and digital twins: Engineers are used to the idea of physics-based modelling. But as AI (machine learning and deep learning) techniques evolve rapidly, the discipline of model-based design continues to evolve. Our framework and product provide a palette to master the landscape and rapidly?acquire and apply new skills in AI and Engineering?to the problem domain.

f)?????I put systems thinking at the top of the diagram so that we get a holistic view of the problem before jumping into the code or the AI

g)????Data driven methods like Kalman estimators are a part of data analytics

h)????Autonomous systems

While this is a complex proposition, its possible to create it incrementally. Hence, the idea of creating an open source Twin builder – by this I mean a workflow engine – based on use cases – that can be built and deployed in a low-code manner – where one physical object or process may have multiple twins - all implemented as Open source.

Welcome thoughts if I have missed anything?

While it is a complex idea, it can be developed incrementally

Finally, the AI for Cyber-Physical systems framework is developed by me as a part of my teaching at the University of Oxford. This is a personal initiative for teaching not associated with any institution. Mistakes are mine alone ??