-
Notifications
You must be signed in to change notification settings - Fork 11
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Merge pull request #159 from w3c/ege-meetup18
Meetup 18 Resources
- Loading branch information
Showing
7 changed files
with
194 additions
and
3 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,189 @@ | ||
Meeting Minutes | ||
=== | ||
|
||
## Info | ||
|
||
:date: **Date:** 23 May 2024 | ||
|
||
### :bust_in_silhouette: Participants | ||
|
||
- Ege Korkan | ||
- Cristiano Aguzzi | ||
- Andrea Giulianelli | ||
- Ivan Dimitry Ribeiro Zyrianoff | ||
- Leonardo Ciabattini | ||
- Marco Di Felice | ||
- Nivya Rachel Cherian | ||
- Lorenzo Gigli | ||
- Luca Sciullo | ||
- Simone Tassi | ||
- Matthew Mustard | ||
- Michael Koster | ||
- Kunihiko Toumura | ||
- Luca Barbato | ||
- Masoud Mohammadgholiha | ||
- Paolo Barbolini | ||
- Tamoghna Ojha | ||
- Jan Romann | ||
- Pontus Pohl | ||
- Fady Salama | ||
|
||
### :scroll: Agenda | ||
|
||
- News from the CG | ||
- Introduction of the Speaker | ||
- Presentation | ||
- Q&A | ||
- Discussion | ||
|
||
:writing_hand: **Scribe:** Ege | ||
|
||
:computer: https://docs.google.com/presentation/d/1qUbYJ6OEp_SSnnXiSUqQJjdpONAfnGh6wE9NhfYsLDY/edit?usp=sharing | ||
|
||
### Introduction | ||
|
||
Cris: Welcome to the 18th meetup! | ||
|
||
### News | ||
|
||
Cris: We don't have another meetup planned, let us know if you have something to present. | ||
|
||
Cris: please join the community group. | ||
|
||
Cris: We also have a small survey. Please fill if you can. | ||
|
||
#### Speaker Introduction | ||
|
||
Cris: Marco Di Felice is a Full Professor at the University of Bologna where he founded the IoT Prism Lab, which focuses on scientific and industrial research in pervasive and mobile systems. He is also faculty member of the international PhD program in Engineering and Information Technology for Structural and Environmental Monitoring and Risk Management (EIT4SEMM) and affiliated to the Advanced Research Center on Electronics Systems (ARCES). He has served as an associate editor for the IEEE Internet of Things journal and the Ad Hoc Networks journal. Currently, he is an associate editor for the Computer Networks journal. He authored more than 160 scientific articles on wireless, pervasive and mobile systems, interoperability and IoT data processing and platforms. | ||
|
||
Cris: Luca Sciullo is a Junior Assistant Professor who authored more than 30 papers in both international conferences and journals. He was a Visiting Researcher at the Huawei European Research Center of Munich, Germany. His research interests include wireless systems, Internet of Things (IoT) Systems, Web of Things (WoT), Digital Twin, Industry 4.0, and Blockchains. | ||
|
||
### Presentation | ||
|
||
Marco: We will present research activity of our lab, focusing on IoT and WoT. | ||
|
||
Marco: I lead IoT Prism Lab. | ||
|
||
> Shows IoT Prism Lab team | ||
Marco: We develop interop to other standards like FIWARE, Arrowhead. We bring WoT to constrained devices, and develop WoT tools. | ||
|
||
Marco: We have a real system and we monitor it. We generate a DT so that we can do stuff like what if analysis, correlation, system state prediction. | ||
|
||
Marco: In the first part, we will focus on monitoring and second part will be digital twin generation and research. | ||
|
||
#### Monitoring | ||
|
||
Marco: We focus today on smart buildings but we did agriculture and healthcare as well. | ||
|
||
Marco: We call this structural health monitoring. It is important in Italy since we have critical infrastructure but a lot of cultural heritage as well. | ||
|
||
Marco: We have tower of Pisa but also one in Bologna as well. | ||
|
||
Marco: We did three projects in this domain. In these, we were asked to create the digital structure. | ||
|
||
Marco: The initial thinking was simple. Put some sensors, get IoT Data, run an AI algorithm to detect. | ||
|
||
Marco: It was not simple. Big data management is difficult to scale to multiple GB. Computation is challenging since we cannot transfer that much data to the cloud and edge does not have enough resources. | ||
|
||
Marco: Training the AI was not easy since we do not have damage at this point. | ||
|
||
Marco: Last challenge was interoperability. There can be different sensors from different manufacturers. Also data processing is done by other entities. | ||
|
||
Marco: So we provide a TD for each sensor. | ||
|
||
Marco: So we came up with a more generic approach. We can abstract from physical things now. | ||
|
||
Marco: We also created MODRON where we adapted WoT Store that we had developed before. You can interact with Things and edit TDs etc. | ||
|
||
Marco: We used it for a vibrating table where a concrete structure is placed on it. | ||
|
||
> Shows the video | ||
Marco: We have simulated an earthquake happened 10 years ago in Birria. | ||
|
||
Marco: Now Luca, the floor is yours. | ||
|
||
#### Digital Twin Generation | ||
|
||
Luca: Since 2022 we are working on generating digital twins, called WoTins. | ||
|
||
Luca: We build a transition between states, using Markov Decision Process and then we can interact with DT through an API like a real Thing. | ||
|
||
> Shows a MDP generated SM. | ||
Luca: MDP states are TD properties. MDP actions are TD actions. | ||
|
||
Luca: By observing the real thing, we can generate the digital twin. | ||
|
||
Luca: We compared our method to: frequency based, machine learning, and hybrid. | ||
|
||
Luca: We used a ground rover which can have unknown obstacles. We evaluted by looking at the grid side and type of estimator. | ||
|
||
Luca: We looked at relativistic digital twins. It considers interacting with twins at different times. | ||
|
||
Luca: When looking at describing the twin, the model can be known in advance. | ||
|
||
> Presents the architecture. | ||
Luca: We had to split the model into readonly and writable variables. | ||
|
||
Luca: The equations can be algeabraic or differential equations. In all cases, the parameters need to be found from real data. | ||
|
||
Luca: we can train the model from different sources | ||
|
||
Luca: We propose to extend the TD with model relevant keywords | ||
|
||
> Shows the terms with the description of values it can take | ||
> Shows a TD snippet with model | ||
> Shows plots with data from digital twin | ||
Luca: We can also do a DT for a drone. We learn the parameters from real data. | ||
|
||
Luca: After some time, the model can predict the real thing very closely. | ||
|
||
Luca: We also look at using WoT in other domains. Also, integrating behavioral model learning with MDP + DL techniques. | ||
|
||
Luca: Also federated Digital Twins is an open topic that we want to explore in the future. | ||
|
||
### Q&A | ||
|
||
Ege: are you using modellica as language for your model expressions? | ||
|
||
Luca: We are using the python syntax. | ||
|
||
Fady: How do you model the action effect on properties? | ||
|
||
Luca: we did a simplification. Writeproperty modifies the property. We need to add such modeling aspects. | ||
|
||
Ege: it would be cool to add how an action input influences the model keyword | ||
|
||
Cris: events are also not part of the models right? | ||
|
||
Luca: yes. We can consider this a start. | ||
|
||
Marco: We can generate the model purely from data or from physical model. | ||
|
||
Luca Barbato: (missed this part due to phone call) | ||
|
||
Luca Barbato: you can submit a use case. | ||
|
||
Luca: I will add the hidden variable problem. External temperature affecting the internal one. | ||
|
||
Marco: This is a real problem. We need to extend in this case. | ||
|
||
Cris: is this information in the td of the physical thing? or is it in the dt td? | ||
|
||
Luca Barbato: it is the first one. We did not want to increase the complexity. | ||
|
||
Andrea: What consists in the digital twin | ||
|
||
Ege: in a previous meetup, we have shown the robot example whose input space is too complex. So the digital twin helps with that | ||
|
||
Marco: a digital twin is a bit different in this case. | ||
|
||
Luca Barbato: You can make a prediction about the future. What will be the future state basically. | ||
|
||
Luca Barbato: In the robot example, you are hiding the complexity. A bit like a validator. |
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Binary file not shown.
Binary file added
BIN
+12.7 MB
Presentations/2024/2024-05-WoTCG-Meetup18-DTs_in_Monitoring-difelice_sciullo.pdf
Binary file not shown.
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters