Urban Digital Twins discussion

[Nicholas]

 

 

[Nicholas]

Welcome to the start of today’s Digital Twin Talk on Urban Digital Twins and a big thank you to @Tanguy Coenen from imec for joining us. We’re looking forward to your thoughts and questions related to Tanguy’s video – and maybe posing one or two of our own.

Please do start adding your thoughts by replying to the conversation thread.
 

[Nicholas]

To get the ball rolling...a first question from me please to @Tanguy Coenen

In your view, what is an Urban Digital Twin?

Thanks, Nick

 

 

[Guest]

Hi Nicholas, 

If a Digital Twin is a virtual replica of a non-virtual entity to support decision making, then an Urban Digital Twin is a virtual replica of a city, to support decision making, related to the city. 

[Michael Mulquin]

The draft definitions of an urban Digital Twin by the IEC (International Electrotechnical Commission) Smart Cities Systems Committee is :

digital representations of a city made up of large quantities of geo-located and other data, often including real time data, which enable better city planning and management.

Note 1: The geo-located data is provided using an integration of Building Information Modeling (BIM) and Geographic Information Systems (GIS).

Note 2: The real-time data is obtained through extensive use of IoT sensors within the city.

Note 3: Developing and using a digital twin of a city involves handling large amounts of data from many sources, which is generally brought together using cloud computing.

Note 4: Artificial intelligence is often used to generate and evaluate different scenarios based on the City Digital Twin to help manage the city better.

 

[Michael Mulquin]

I'd very much value comments and suggestions of how to improve this.

 

[Guest Hello Mr Coenen,]

Thank you very much for your webcast and introduction to BuDI - My question relates to your YTT - looking at yesterday, today, tomorrow- the data for these time frames is very different by nature. It is precisely where we at BAIKAL AR are working. How do you think that machine learning can deal with the huge differences in data types & formats, quantity, errors, gaps, - for example manually observed data such as temperature, air quality or numbers of people, to early electro mechanical data in road and traffic lights,  door counters, and now the plethora of electronic IOT devices. The more recent data will swamp the historical, will it not? 

[Guest]

Hi Michael,

Thank you for your insights!

As I mention in the video, we try to keep up-to-date with current international efforts on smart cities, so we would be very interested to engage in conversation on the topics you mention.

If there are any references to texts you could send over that provide more info, that would be great too.

[Guest Charles Keen]

Name was omitted

[DRossiter87]

Hi Tanguy, 
Thank you for the video, very interesting.  You mentioned the need to manage the latency between the changes to the physical asset and its representation digitally.  What I wanted to ask if there are two distinct latencies:

1. The time from an intervention to when it is reflected physically (e.g. a planned intervention such as a road closure);
2. Time time from when the sensor data is collected to when it is reflected digitally (e.g.  to import, configure, review, and publish)

Please see image below (trying) to explain what I mean

Edited May 12, 2020 by DRossiter87

[Nicholas]

3 minutes ago, Tanguy Coenen said:

Hi Michael,

Thank you for your insights!

As I mention in the video, we try to keep up-to-date with current international efforts on smart cities, so we would be very interested to engage in conversation on the topics you mention.

If there are any references to texts you could send over that provide more info, that would be great too.

Hi @Michael Mulquin

Many thanks for your comments. We see a lot of interest from local authorities and cities in digital twins and it would be great to explore the links between smart cities and digital twins.

@Tanguy Coenen it would also be great if you have any thoughts on the difference or similarities between the two please.

Many thanks

Nick

 

 

 

[Michael Mulquin]

In the IEC we are collecting and analysing use cases for City Information Modeling/urban digital twins in order to define the standards that are needed.  As a Systems Committee within the IEC, our role is not mainly to develop standards, but to take a systems approach to understand all the different kinds of things needed for a new technology to be successful - data standards, IoT standards, communications standards, management standards, etc. etc. and then provide the information to the relevant Technical Committees in the appropriate Standards Development Organisations to help them develop the family of standards needed.  We formally started work on this project in January, but things have been put on hold for a while, due to Covid 19. We hope to restart things soon.

We very much welcome experts to join us in this, so please get in touch, if you would like to find out more.

IEC_TS_63273_ED1_City_Information_Modeling_Use_Case_Collection_20191211.pdf

[Nicholas]

3 minutes ago, DRossiter87 said:

Hi Tanguy, 
Thank you for the video, very interesting.  You mentioned the need to manage the latency between the changes to the physical asset and its representation digitally.  What I wanted to ask if there are two distinct latencies:

1. The time from an intervention to when it is reflected physically (e.g. a planned intervention such as a road closure);
2. Time time from when the sensor data is collected to when it is reflected digitally (e.g.  to import, configure, review, and publish)

Please see image below (trying) to explain what I mean

Hi @DRossiter87

A quick follow up  while Tanguy is also having a think about this.

A lot of the DT Hub members have talked about a need for digital twins, and the data that feeds into them, to be "right time" rather than real time. This could vary considerably from say, long term planning related to requirements for housing, vs managing incidents that impact on public safety (where real time, or near real time data might be required).

@Michael Mulquin and others it would be great if you also have any particular thoughts on this please.

Many thanks

Nick

 

 

 

[Guest]

7 minutes ago, Guest Hello Mr Coenen, said:

Thank you very much for your webcast and introduction to BuDI - My question relates to your YTT - looking at yesterday, today, tomorrow- the data for these time frames is very different by nature. It is precisely where we at BAIKAL AR are working. How do you think that machine learning can deal with the huge differences in data types & formats, quantity, errors, gaps, - for example manually observed data such as temperature, air quality or numbers of people, to early electro mechanical data in road and traffic lights,  door counters, and now the plethora of electronic IOT devices. The more recent data will swamp the historical, will it not? 

Hi Charles,

With historical data, I was referring to data measured in the same system, with the same sensors, but a certain time interval in the past. Are you also looking at it in this way? I'm not sure you are, as you talk about "early electro mechanical data in road and traffic lights,  door counters, and now the plethora of electronic IOT devices". Maybe you are interpreting "historical data" here as "older ways to measure things"?

In any case, we are currently applying ML to Urban Digital Twin problems like interpolating air quality measurements in cities. Where measurements are sparse, we use ML techniques to make sure we can build a heat map of a city. This to me is a way to apply ML to improve datasets, but I'm not sure it answers your question. If not, could you please elaborate?

[Guest Luke O'Rafferty]

Hi Tanguy,

Interesting to see what you are doing with Digital Twins.

You talked about the importance of us Machine Learning over manually driven processes. Are you getting enough engagement/access to "feed the machine" and enable it to realy bare fruit?

[Andrew Myers]

Thanks for sharing - this was a really interesting video. It was good to see the importance of open standards and interoperability being stressed, this is going to be key to ensuring that investment in twins today doesn't become obsolete as technology moves on.

I'm keen to know what stage you are at with these projects - do you have any larger multi-domain twins in use today, or are most still in the proof of concept stage?

[Guest]

2 minutes ago, Guest Luke O'Rafferty said:

Hi Tanguy,

Interesting to see what you are doing with Digital Twins.

You talked about the importance of us Machine Learning over manually driven processes. Are you getting enough engagement/access to "feed the machine" and enable it to realy bare fruit?

Hi Luke,

Feeding the machine has been a big part of our efforts in recent times. As imec comes at Smart Cities mainly from an IoT perspective, working with sensors and IoT is at the core of our expertise. For example, we've been doing research on new ways to measure/map air quality using mobile sensors and machine learning. It's still research, but with promising results. Another example is doing sensor fusion research on people flows in cities. Here, we're working with multiple partners (e.g. telco's) who provide us with data sources, but we're still working on validating and learning from these models. Finally, we also have been looking at active learning as a way to do semi-supervised learning in smart city environments. 

[Nicholas]

6 minutes ago, Michael Mulquin said:

In the IEC we are collecting and analysing use cases for City Information Modeling/urban digital twins in order to define the standards that are needed.  As a Systems Committee within the IEC, our role is not mainly to develop standards, but to take a systems approach to understand all the different kinds of things needed for a new technology to be successful - data standards, IoT standards, communications standards, management standards, etc. etc. and then provide the information to the relevant Technical Committees in the appropriate Standards Development Organisations to help them develop the family of standards needed.  We formally started work on this project in January, but things have been put on hold for a while, due to Covid 19. We hope to restart things soon.

We very much welcome experts to join us in this, so please get in touch, if you would like to find out more.

IEC_TS_63273_ED1_City_Information_Modeling_Use_Case_Collection_20191211.pdf 3.62 MB · 1 download

Hi @Michael Mulquin

Many thanks for sharing this. It would be great to learn more about the use case information that you are collecting and to work out how best to share some of that thinking on the DT Hub (we'll take a look at the attachment you shared to start with).

Also, please can you let this community know the best way to get involved in the project you mention.

Thanks again, Nick

 

 

 

[Michael Mulquin]

6 minutes ago, Nicholas said:

Hi @DRossiter87

A quick follow up  while Tanguy is also having a think about this.

A lot of the DT Hub members have talked about a need for digital twins, and the data that feeds into them, to be "right time" rather than real time. This could vary considerably from say, long term planning related to requirements for housing, vs managing incidents that impact on public safety (where real time, or near real time data might be required).

@Michael Mulquin and others it would be great if you also have any particular thoughts on this please.

Many thanks

Nick

 

 

 

I agree that "right time" is better. The contrast is really between static data, where the focus is not time related at all - data about the physical buildings and urban environment, about general demographics etc, and data about how things change over time. If you are wanting to monitor change - either to react to what is happening or to look at scenarios about what changes might happen, then you need data at the time intervals that are important.  I guess that is one area where we need to change our draft definition!

[Guest]

1 minute ago, Andrew Myers said:

Thanks for sharing - this was a really interesting video. It was good to see the importance of open standards and interoperability being stressed, this is going to be key to ensuring that investment in twins today doesn't become obsolete as technology moves on.

I'm keen to know what stage you are at with these projects - do you have any larger multi-domain twins in use today, or are most still in the proof of concept stage?

Hi Andrew,

I'm glad you liked the video.

Yes, open standards and interoperability are very central to everything we do in the Smart City domain. Imec is also advising the Flemish government on how to build open IoT smart city architectures, but that's another project. 
We are still in proof-of-concept mode and are working with Flemish cities to have BuDi in operational use around the end of 2020, beginning of 2021. Also, we're  working on implementations in the "private built environment" which should go online end of 2020. FInally, there's the Duet H2020 project (https://www.digitalurbantwins.com/), which will normally see implementations in the course of 2021. 

[Nicholas]

Another question - for @Tanguy Coenen and all.

What are the most valuable/attractive use cases for digital twins in cities? 

(Maybe some of these might add more examples to help with the right time vs real time debate too)

Many thanks,

Nick

 

[Guest]

 

9 minutes ago, Michael Mulquin said:

I agree that "right time" is better. The contrast is really between static data, where the focus is not time related at all - data about the physical buildings and urban environment, about general demographics etc, and data about how things change over time. If you are wanting to monitor change - either to react to what is happening or to look at scenarios about what changes might happen, then you need data at the time intervals that are important.  I guess that is one area where we need to change our draft definition!

Hi Michael and Dan,

Yes, I agree that real-time is a very relative concept which depends on the type of decision you want to support. So indeed, right-time may be a better term.

[DRossiter87]

Just now, Nicholas said:

Another question - for @Tanguy Coenen and all.

What are the most valuable/attractive use cases for digital twins in cities? 

(Maybe some of these might add more examples to help with the right time vs real time debate too)

Many thanks,

Nick

 

Hi Nick,

Yes, we recently hosted a webinar on Value where we looked at possible use cases for digital twins.  What seemed to resonate well as the idea that use cases was the application of "people", "planet" and "profit" as domains for each use case. 

This also seems to have parallels with the Flourishing systems report (published this morning): https://www.cdbb.cam.ac.uk/files/flourishing-systems_final_digital.pdf  which focuses on economical infrastructure, social infrastructure, and natural environment.

[Guest]

5 minutes ago, Nicholas said:

Another question - for @Tanguy Coenen and all.

What are the most valuable/attractive use cases for digital twins in cities? 

(Maybe some of these might add more examples to help with the right time vs real time debate too)

Many thanks,

Nick

 

Hi Nicholas,

There are many and the value of a use case highly depends on the local challenges of a city, as no two cities are the same.
Yet, we see a lot of traction in these domains:
a.    People flows: where do people go to in the city, either in motorized (cars, trucks,…) or non-motorized way (bicycles, on foot,…)
b.    Air quality: what does the air quality (particulate matter, CO2, NO2, Ozone) in a city look like and how is it impacted by certain interventions, e.g. in the traffic domain?
c.    Flooding: what areas in the city are likely to flood in a city, giving the current weather predictions, and what can we do about it?
 

[Michael Mulquin]

12 minutes ago, Nicholas said:

Hi @Michael Mulquin

Many thanks for sharing this. It would be great to learn more about the use case information that you are collecting and to work out how best to share some of that thinking on the DT Hub (we'll take a look at the attachment you shared to start with).

Also, please can you let this community know the best way to get involved in the project you mention.

Thanks again, Nick

 

 

 

The IEC is a sister organisation of ISO and we have the same normal process for people to get involved, which is via their national Standards organisation - in the UK it would be via BSI. However, for our systems work we are more flexible. Systems Committees in the IEC can offer "R" membership to interested bodies, which enables them to take a full role in our work. We also can utilise a "pool of experts" to provide us with individual comments and suggestions. If anyone would like to find out more - let me know.