186 - Egressibility: a paradigm shift in evacuation research with Enrico Ronchi

Transcript

WEBVTT

00:00:00.980 --> 00:00:02.688
Hello everybody, welcome to the Fire Science Show.

00:00:02.688 --> 00:00:07.711
There is this buzzword, paradigm shift that a lot of people are using.

00:00:07.711 --> 00:00:14.670
I'm also using that word quite a lot and when I hear it it usually triggers me, and I know it triggers a lot of people.

00:00:14.670 --> 00:00:25.951
As someone who really enjoys Thomas Kuhn's philosophy and I've read the Scientific Revolution's book at least three to four times I'm very mindful about where I place such a big word.

00:00:25.951 --> 00:00:43.067
Paradigm shift means changing everything we know, changing the approach when your ordinary science does not work anymore, making such a shift that everything changes and suddenly you can accommodate that new paradigm that did not work with the previous science.

00:00:43.067 --> 00:00:56.960
And while it's a buzzword and it triggers me, sometimes you get to see science which is a true paradigm shift, which really has a true paradigm shifting potential, and that's the type of research we're talking about in this podcast episode.

00:00:56.960 --> 00:01:12.250
So ERC grants were given out in December Consolidator, two grants were given to topics around fire, which is astounding, like we've never had two grants given out to our field in the history.

00:01:12.250 --> 00:01:13.772
So that's amazing.

00:01:13.772 --> 00:01:19.841
And today I'm talking with one of those grantees, with my colleague, professor Enrico Ronchi from Lund University.

00:01:19.841 --> 00:01:21.828
You've heard Enrico on this podcast multiple times.

00:01:21.828 --> 00:01:31.147
He's been a guest since the start of the podcast and he's one of been a guest since the start of the podcast and he's one of the main figures in the world of evacuation science and human behavior in fires nowadays.

00:01:31.147 --> 00:01:41.933
And he just applied for a crazy grant that aims on paradigm shifting in the whole field of evacuation, with those at biggest disadvantage in mind.

00:01:41.933 --> 00:01:58.912
The grant is called Egressibility and Enrico is really trying to change the way how we understand the evacuation of the population at disadvantage disabled, elderly and figure out new ways to help them in buildings.

00:01:58.912 --> 00:02:13.411
That includes a lot of new research, a lot of new experiments, use of innovative tools and really flipping the perspective, like looking more on the disabilities themselves, looking more at human capabilities and linking those to the evacuation process.

00:02:13.411 --> 00:02:19.871
It's really interesting In this discussion, in this podcast episode, I've tried to pull Enrico about.

00:02:20.032 --> 00:02:20.933
What do we know today?

00:02:20.933 --> 00:02:24.069
So I want this episode also to be directly useful to you.

00:02:24.069 --> 00:02:32.289
So I'm really trying to find the edge of our knowledge today and where Enrico is trying to push that edge.

00:02:32.289 --> 00:02:41.546
So from the episode, you'll not just learn about the amazing project that Enrico just got, but you'll learn a lot about where we are today in modeling evacuation.

00:02:41.546 --> 00:02:45.165
So I know it's a nice journey, so stay till the end.

00:02:45.165 --> 00:02:48.056
At the end, enrico tells you the secret how he got the grant.

00:02:48.056 --> 00:02:48.599
That's hilarious.

00:02:48.599 --> 00:02:52.151
Anyway, let's spin the intro and jump into the episode.

00:02:56.981 --> 00:02:58.622
Welcome to the fire science show.

00:02:58.622 --> 00:03:25.157
My name is Wojciech Wigrzyinsky and I will be your host Consultants, a multi-award-winning independent consultancy dedicated to addressing fire safety challenges.

00:03:25.157 --> 00:03:36.846
Established in the UK in 2016 as a startup business of two highly experienced fire engineering consultants, the business has grown phenomenally to eight offices across the country, from Edinburgh to Bath.

00:03:36.846 --> 00:03:46.354
Colleagues are on a mission to continually explore the challenges that fire creates for clients and society, applying the best research, experience and diligence for effective, tailored solution.

00:03:46.354 --> 00:03:50.506
In 2025, there will be new opportunities to work with OFR.

00:03:50.506 --> 00:03:58.187
Ofr will grow its team once more and is keen to hear from industry professionals who would like to collaborate on FHIR safety features this year.

00:03:58.187 --> 00:04:00.927
Get in touch at ofrconsultantscom.

00:04:00.927 --> 00:04:05.829
Hello, I'm joined here today by Professor Enrico Ronchi from Lund University.

00:04:05.829 --> 00:04:06.902
Hello, enrico.

00:04:07.424 --> 00:04:09.711
Hi Wojciech, Very nice to be back.

00:04:10.252 --> 00:04:11.937
And what a great circumstance.

00:04:11.937 --> 00:04:15.408
Congratulations on your ERC Consolidator Grant.

00:04:15.689 --> 00:04:16.713
Thanks, thanks, wojciech.

00:04:16.713 --> 00:04:26.173
I'm so happy to be here talking about this because, like you heard in the previous ERC episodes, it is so much work to get to this point.

00:04:27.194 --> 00:04:31.951
I know, and let's say, the Fireside Show interview is the grand finale of it.

00:04:31.951 --> 00:04:36.992
It's just pure pleasure at this point after all those tears and work, but in fact it is.

00:04:36.992 --> 00:04:44.471
It is a massive achievement and it's a consolidated grant, so it's the middle tier for mid-career researchers.

00:04:44.471 --> 00:04:47.728
So that's even more competitive scheme than the starting grant.

00:04:47.728 --> 00:04:49.447
So especially amazing.

00:04:49.447 --> 00:04:56.454
And of course you're an evacuation scientist and your subject of research is tied to evacuation.

00:04:56.454 --> 00:05:01.591
The project is called egressibility, a paradigm shift in evacuation research.

00:05:01.591 --> 00:05:04.449
And let's start with egressibility.

00:05:04.449 --> 00:05:06.161
What the hell does egressibility mean?

00:05:07.084 --> 00:05:12.237
Yes, I mean the idea is to merge the words of evacuation and accessibility.

00:05:12.237 --> 00:05:23.490
So the definition of egressibility is accessibility to means of evacuation, and this is an idea that I've been boiling down for quite some time.

00:05:23.490 --> 00:05:45.490
I had an earlier project here in Sweden, together with some colleagues from the medical faculty that are working more in the accessibility space, and we started talking about how can we characterize people with disabilities and different type of functional limitations so that we can get this information and make it useful for evacuation design.

00:05:45.490 --> 00:05:59.649
And then, you know, we went back into the literature and I came across this keyword from Ghislaine Proulx, that is, an historical researcher in the world of human behavior in fire, which, as you probably know, also passed away too early.

00:06:00.560 --> 00:06:12.175
And then I say, okay, my only chance to get an ERC is to make a very clear case that I'm doing something completely new and addressing a very relevant problem.

00:06:12.175 --> 00:06:29.387
And that's why I thought, okay, we have aging populations and we know everything that is happening around us related to climate change at bigger scales, but also at BIMD, and everything that has to do with geopolitical uncertainty, security, everything that can trigger an evacuation.

00:06:29.387 --> 00:06:40.740
And then I said, okay, I'm going to bring the concept of evacuation to NERC through egressibility, so through looking at what are the vulnerable populations.

00:06:40.740 --> 00:06:43.509
So that's a bit the general idea of the project.

00:06:44.220 --> 00:07:02.107
So it would be something like not just what evacuation time you have or not what average walking speed person will take to go through an evacuation pathway, but more in how well this evacuation tool set can serve the population, including those that could be at disadvantage.

00:07:02.468 --> 00:07:05.694
Yeah, I mean, the idea is to flip the coin.

00:07:05.694 --> 00:07:15.428
I mean, nowadays what we do is that we design buildings for the average person and then we do dedicated solutions for people that have different types of disabilities.

00:07:15.428 --> 00:07:25.026
Right, there has been a lot of discussion into using elevators for people with disabilities or, I don't know, different special types of alarms and things like this.

00:07:25.026 --> 00:07:57.110
But in order to flip the coin, we should start looking at a completely different population when we do experimental work or when we study case studies, which is a bit of the opposite of what is happening nowadays, because, you know, the biggest pool of people that we do experiments with nowadays is students, because that's the one that we have easier access to, and those are young, generally, mostly healthy, and this is completely on the opposite end of what are the most vulnerable population in a vacation.

00:07:57.129 --> 00:08:41.735
So flipping the coin means looking at a completely different pool of populations to begin with, especially going into older populations, where we know that, you know, all of us, unfortunately, before or after in our life, we start having a decline in our functional abilities, which means that this literally affects everyone, and that's been my strongest argument for the ERC that this is a problem that really concerns everyone and you know, if we look at different emergency scenarios, fires especially, and if you look at the statistic of who dies in a fire, I mean it's pretty obvious that the great majority of those that suffer the most are the ones that have some sort of functional limitation very often.

00:08:41.735 --> 00:08:44.745
I mean even the cases that we read now in the news.

00:08:44.745 --> 00:08:55.184
I mean you heard really heartbreaking stories now from the LA wildfires, but I mean we heard this in Grenfell before with people with disabilities.

00:08:55.184 --> 00:09:00.221
I mean we hear it all the time that the most affected are always the most vulnerable.

00:09:00.480 --> 00:09:15.571
Yeah, I absolutely agree with that, and I also had an episode in the Fire Science Show with Marie Button where we talked about evacuation from a perspective of a disabled fire engineer so she's a wheelchair user and you didn't listen to that, dear listener.

00:09:15.571 --> 00:09:28.267
I would recommend that to open your mind to the problem.

00:09:28.267 --> 00:09:30.852
Anyway, I want to talk about the.

00:09:30.852 --> 00:09:33.865
I'm fascinated by people who get ERC grants.

00:09:33.865 --> 00:09:35.269
That used to be my dream.

00:09:35.269 --> 00:09:37.241
Perhaps it will be become a dream again.

00:09:37.241 --> 00:09:44.614
Anyway, going back to your grant, but it goes much further beyond just finding a new fundamental diagram just for old people, right?

00:09:44.614 --> 00:09:46.302
It's not about that, right?

00:09:46.663 --> 00:09:48.330
No, it's not really about that.

00:09:48.330 --> 00:09:51.280
So it's, and it's not just because that's the other thing.

00:09:51.280 --> 00:10:07.735
When we think about people with disabilities, by default we kind of associate an image of a person in a wheelchair, but that's just one of the multiple you know possible functional limitations that you may have, the mobility ones.

00:10:07.735 --> 00:10:20.033
So the idea is to work more on characterizing what are, you know, all the abilities of people and how those can decline over time and what impact they have on evacuation.

00:10:20.033 --> 00:10:28.883
Because, again, I really learned a lot by working with people in the medical world and using all this classification that I use in the health science domain.

00:10:28.923 --> 00:10:53.697
Especially the WHO has a very nice classification which is called ICF the International Classification of Functioning and Disability and Health, which basically makes a very clear categorizations on all the different abilities and functions that our body has and how those can be affected either by a disability or by our body decline with the age.

00:10:53.837 --> 00:11:02.081
So basically, our often, with age, we have parts of our body that start declining in functioning, and that's the idea.

00:11:02.582 --> 00:11:14.394
So to use this information from the health science to learn more about people and carry this information into the world of evacuation.

00:11:14.394 --> 00:11:28.235
So how we can use the information that we have in the characterization of the population to understand more of decisions, because you know, it's not black and white, it's not like I'm on a wheelchair or I'm not in a wheelchair.

00:11:28.235 --> 00:11:35.106
There is a lot of shades of gray on what kind of functional limitations you have and how they can affect your evocation.

00:11:35.106 --> 00:11:51.399
So the idea is to then link those with decision-making, so how you know if you have issues in balance when you walk down the stair because you have reached a certain age, how this affects your decision to even take a stair or to remain at home during a fire.

00:11:51.399 --> 00:11:52.785
So this is a bit of the idea.

00:11:52.785 --> 00:12:02.575
So to link decision making with a much more detailed characterization of your functional abilities, which will have to be borrowed by the health science domain.

00:12:02.575 --> 00:12:05.669
So I really need to look into that literature.

00:12:06.080 --> 00:12:13.427
Yeah, so that would also cover things like potential difference in, let's say, predistribution time, like you're talking about paradigm shift.

00:12:13.427 --> 00:12:25.226
So I assume at the end you're going to dump the pre-evacuation time paradigm and just replace it with something bigger, or I assume your thing is a bigger thing, but with the current engineering it would be.

00:12:25.226 --> 00:12:28.793
The response to the fire will be different when you are disabled.

00:12:28.793 --> 00:12:31.783
Uh, perhaps the preparedness even would be different, right?

00:12:32.567 --> 00:12:57.549
I mean the ambition and I know this is very ambitious, and that's the whole point of the of the erc is to basically try to move towards a completely different types of models that are not like the classic agent-based type of models that we see nowadays used in evacuation simulations, but moving towards pure data-driven models, like pure machine learning-based.

00:12:58.130 --> 00:13:09.082
And the way I call this is inclusive, because there is a lot into looking, looking at machine learning models, but nothing on inclusive machine learning.

00:13:09.082 --> 00:13:12.052
So how can we make sure that we don't have representation bias?

00:13:12.052 --> 00:13:30.634
Because, again, if you build a data driven model on a population of students, you're gonna spit out a result which is representing the decisions and behavior of students, but if you have a much wider population, you should be able to predict a much wider outcome.

00:13:30.634 --> 00:13:55.711
So that's the idea to move from classical agent-based models to data-driven, inclusive models which basically spit out some sort of a probability of taking a given decision based on the environmental conditions, all those classic factors that we look at in the evacuation models, along with a much more detailed characterization of your health, basically, or your functional abilities.

00:13:56.600 --> 00:14:08.528
So if you had to pick up the main parts on your pathway to get there, what do you need to do from day one, which starts soon, I guess to the last day of your grant?

00:14:08.528 --> 00:14:09.985
What's going to happen in between?

00:14:09.985 --> 00:14:11.028
How do you plan it out?

00:14:11.921 --> 00:14:14.460
I think that there are two avenues here.

00:14:14.460 --> 00:14:36.208
The first one is to really learn more and more from the health science domain into how we can characterize people, because, you know, in engineering I would say, of course we are not medical doctors, so we need to use very simplified assumptions and so on, but in research we don't have to, you know, we can go back to fundamentals and we can talk.

00:14:36.208 --> 00:14:48.177
You know, I have a very nice pool of colleagues and collaborators here in the medical faculty at Lund University I work with on this topic and we can all learn in the fire engineering ward into a better characterization of people.

00:14:48.177 --> 00:14:54.991
And that's the first track, so to understand more on who each person that is involved in a vacation is.

00:14:54.991 --> 00:15:12.452
And the second avenue is to then design scenarios and design data collection methods that relate to decision making, in which we can understand what are the implications on your decision making of your functional abilities.

00:15:12.452 --> 00:15:18.051
And that's why I'm looking at a very diverse pool of research methods to use.

00:15:18.130 --> 00:15:42.876
So from VR which you know I am a big enthusiast and a very big user for some years but also to talk to people, to do interviews and also review case studies Because, as I said, there has been a lot of events in which, very commonly, we can see that the most affected population are those with disabilities, are those with disabilities.

00:15:42.876 --> 00:15:55.666
So there is already a lot of real-world decisions that we can clearly see indicating that there are wide differences in decision-making process depending on what kind of functional limitations or disabilities you have.

00:15:55.666 --> 00:16:09.565
So that's the idea both designing a variety of data collection methods to understand more about decision-making depending on who you are, and a better characterization of who you are.

00:16:09.565 --> 00:16:21.942
So try to have a much deeper screening again, trying to learn from the health sciences, which is a challenge, because I mean me and many people that listen to your podcast.

00:16:21.942 --> 00:16:24.067
We are engineers, so we don't come from that board.

00:16:24.106 --> 00:16:27.854
But I mean, no one prevents us to learn, you know, especially when we do research.

00:16:28.461 --> 00:16:37.732
I'll ask you a difficult question, but perhaps it would be of great value to the listeners, who could use this information to process whatever they're doing at this time, not five years from here.

00:16:37.732 --> 00:16:58.557
So let's imagine if you would like to do exactly what you said, but with the state of knowledge that is today, like no further research, what would you say would be the bottlenecks or the edge of our understanding at the level of the population, the decision making and the general evacuation process?

00:16:58.557 --> 00:17:02.148
Let's start with the population, like where's the end of our knowledge today?

00:17:03.581 --> 00:17:14.151
First of all, when we characterize samples, when we do experiments, for instance, with VR or whatever other type of data collection, the characterization of the sample that we do is very crude in engineering.

00:17:14.151 --> 00:17:18.210
So, okay, we have male or female, or you know, you have….

00:17:18.530 --> 00:17:19.271
Sometimes age right.

00:17:20.644 --> 00:17:38.111
Age and things like this, but we don't have, like you know, the this classification that I mentioned goes down to the level of to which extent I'm open, I'm able to operate something, to which extent my limbs function, how my stamina is when I'm doing a given task.

00:17:38.111 --> 00:17:49.671
So it's much more level of depth into how our functional abilities in daily life and, in our case, will have to be translated to emergencies are working.

00:17:49.671 --> 00:17:56.953
So, first of all, we need to really learn into characterizing in much more depth the populations that we study.

00:17:57.221 --> 00:18:03.614
So you would need a much bigger granularity in the description of the data sets that you already work with.

00:18:03.614 --> 00:18:05.816
That's the struggle Exactly exactly the description of the data sets that you already work with.

00:18:05.836 --> 00:18:10.146
That's a stop-bomb indeed, exactly because the data that we have are much more crude when it comes to describing people.

00:18:10.146 --> 00:18:23.846
I mean, yes, age, gender, a couple of other key variables, but that's pretty much it when it comes to health, while instead there is many more variables that are much more refined that can play a role.

00:18:23.866 --> 00:18:37.998
Yeah, Sorry I'm breaking your thought chain, but this is really interesting because you know, sometimes you would have those massive discrepancies in the walking speeds or something like, let's say, from one meter to two meters per second, like 100% distribution.

00:18:37.998 --> 00:18:41.199
And your data said there are people 20 to 60, right?

00:18:41.199 --> 00:18:45.300
So perhaps if you had granularity it would explain the differences.

00:18:45.300 --> 00:18:52.324
So, indeed, looking at the data with more knowledge of what came into that data can really open new pathways.

00:18:52.324 --> 00:18:56.494
But of course you have to repeat some studies because the ones that are today.

00:18:56.875 --> 00:19:14.519
with just this data, it's not that you can magically increase the depth resolution of those right and you know, and there is also, of course that's one of the factors, because it's also not just about you know, your, because your physical ability describes this your so-called upper limit, let's say of what you could do.

00:19:15.122 --> 00:19:18.797
But of course there is a lot about, uh, you know, decision making, motivation.

00:19:18.797 --> 00:19:21.962
I mean, we did, uh, several experiments over time.

00:19:21.962 --> 00:19:50.311
There are a few research studies that look like there are very vast differences in walking speed just because people have different motivations, right, I always recall I do an example of these experiments that were done here in Sweden in this project that with some colleagues we had on ascending evacuation and basically we could see that it was a very strenuous task to go upwards and we could see a boost in walking speeds towards the end of the task because people could see the end of the goal.

00:19:50.792 --> 00:19:57.779
It's like when you run a marathon, you know the last lap is probably the one that you go fastest because you know it's done, you made it so.

00:19:57.779 --> 00:20:19.453
So there is like that side of the equation which is more like on motivation and the decision-making drivers, but on the other hand, we have very limited understanding on our upper limits of what we can actually do and also how aware we are of our upper limits and how this influence our decision-making and evacuation.

00:20:19.453 --> 00:20:34.376
Because you know, the bottom line is also some people may take a given decision simply because they're not confident enough with their abilities to take a given path or to to take a given, let's say, decision and how will affect down the line.

00:20:34.376 --> 00:20:37.932
There's likelihood of surviving a fire and you.

00:20:37.991 --> 00:20:40.479
You wanted to mention the second thing about population.

00:20:40.499 --> 00:20:47.161
Yeah, yeah, the second thing, which is about you know what is the big hinder that we have today is the amount of data that we have.

00:20:47.201 --> 00:20:54.019
So one thing that I want to do also is more to try to scale up the quantity of data that we can collect.

00:20:54.019 --> 00:21:07.829
And, you know, and this can be, that's why I'm aiming at different types of data collection methods vr, you know it allows you to collect quite large amount of data in a relatively shorter time, but also deploy this online.

00:21:07.829 --> 00:21:33.461
Because, you know, we worked in a project a couple of years ago with Ann Templeton and a few others were involved, john Drury and we started doing VR online and, you know, this kind of interactive behavioral intention experiment, so not just a crude questionnaire, but something that is kind of in between a classic questionnaire and a VR that you can deploy online.

00:21:33.461 --> 00:21:37.260
And there, you know, there are platforms that help you to collect data.

00:21:37.260 --> 00:21:45.917
And if you're able to design those in an inclusive way because that's the other challenge If you want to target the most varied group, you cannot.

00:21:45.998 --> 00:21:49.825
You know, you open a Pandora's box linked to accessibility.

00:21:49.825 --> 00:22:03.434
You know I started thinking about audio games, all sorts of things that are not just classic visual stimuli, so, or all sorts of other triggers that you can have and scenarios that you can have to understand the decision-making and evacuation.

00:22:03.434 --> 00:22:17.798
But if you are able to decide this, then you can have and scenarios that you can have to understand the decision making and evacuation, but if you are able to decide this, then you can scale it up, because if you deploy things online, then you can collect thousands, tens of thousands of data and that really helps to have a much larger database.

00:22:17.877 --> 00:22:24.821
And if you connect that with a good screening of who they're doing that that kind of test, then you can get a very good pool of data.

00:22:25.609 --> 00:22:30.001
Like your rough estimation of how much data we have today and how much more you would need.

00:22:30.851 --> 00:22:43.650
We talk about ideally, we will need from 10 to 100 times more data than we have today at least to start working with data-driven models, and you know I've seen this done, for instance, in the world of crowd dynamics.

00:22:43.650 --> 00:22:51.655
You know that I'm very close to the pedestrian and vacation dynamics community and, for instance, they start collecting this type of data at very large scale.

00:22:51.655 --> 00:22:57.059
You know million of trajectories from regular normal.

00:22:57.059 --> 00:23:07.326
You know walking not from emergencies, and you know for emergencies it's much harder to do this unless you do it in a controlled way, in an experimental setup, because these are rare events.

00:23:07.326 --> 00:23:14.313
You know we don't have an emergency every day, while instead, if you have to collect data from a train station, you can do it every day for a year.

00:23:14.373 --> 00:23:17.521
That's what they did these colleagues in Eindhoven and University of Eindhoven.

00:23:17.521 --> 00:23:26.798
They basically collect trajectory of people for one year, every day, and they got a very large pool of trajectories and then they start understanding the patterns of movements.

00:23:26.798 --> 00:23:35.800
But, as I said, we can mirror this kind of modeling approaches which are data-driven, but for doing so we need a much larger pool of data.

00:23:35.800 --> 00:23:49.109
I mean, we need to talk about tens of thousands, hundreds of thousands of data points to be able to build something that is robust enough, especially when you want to have a wide variety of representativeness in the population.

00:23:49.109 --> 00:23:51.719
You cannot just have the regular students.

00:23:52.210 --> 00:24:14.623
But it's good that there are already examples of scaling up this type of data accumulation and I think your idea of scaling up through also those gamified interfaces, virtual immense realities and all the tool sets that make this much easier than chasing people through a tunnel filled with smoke, like you was kind enough to do some years ago, it sounds much easier.

00:24:14.623 --> 00:24:25.124
Decision making when are we today with understanding the decision making process and how much you need to know more to turn aggressability into reality?

00:24:25.711 --> 00:24:34.355
I think we have very good theories that help us understanding some of the fundamental concepts of human decision-making in fire emergencies.

00:24:35.356 --> 00:24:39.484
What we need is to have a more refined understanding.

00:24:39.484 --> 00:25:12.123
So try to link more the data that we have to the type of people that we have, because these are general theories that will explain again what the great majority of people will do and, you know, give us overall patterns, are the ones that don't really necessarily behave according to the book, because maybe they are not able to take a staircase or because they cannot really easily get information and so on.

00:25:12.123 --> 00:25:24.119
So, and that's the thing, try to have a wider understanding on decision-making, not just for the average person, but even targeting specific groups.

00:25:24.119 --> 00:25:34.768
And again, there I my goal, and again this is also very ambition, but it's try to be systematic in characterizing populations that have different types of disabilities, for instance.

00:25:34.768 --> 00:25:48.390
So design experiments for someone that is visually impaired, or design an experiment for someone that has an hearing impairment, and so on, and target those groups and understand how their decision-making process works, because on this we know very little.

00:25:48.611 --> 00:25:51.559
And because it has to be a part of a model, the decision-making process.

00:25:51.559 --> 00:26:02.993
Are you thinking more like Erika Kulgoski-style PADM-type models where there's a decision tree, or you're thinking more about just the distribution of outcomes and assigning probabilities?

00:26:02.993 --> 00:26:03.335
I don't know?

00:26:03.335 --> 00:26:08.670
Monte Carlo-ing the outcomes or another branch like machine learning and just having a blackboard?

00:26:08.710 --> 00:26:15.596
This will have to be probabilistic but, I will basically work with machine learning model, because when you scale up data.

00:26:16.711 --> 00:26:32.691
I think we can go that path to work with data-driven approaches, and the idea will be indeed to have on one side the characterization of the persons and on the other side the probability of taking a decision and trying to see if you can find good correlations.

00:26:32.691 --> 00:26:38.411
And again, the good thing with ERC which is good and bad in a way is that it gives you a lot of freedom.

00:26:38.411 --> 00:26:52.217
So I laid down a general description of the modeling approach that I want to take, thinking about this inclusive machine learning and thinking about, for instance, decision trees, the way they work with machine learning.

00:26:52.217 --> 00:26:58.897
They seem very well fit to work with evacuation decisions and I've seen some initial application in our world.

00:26:58.897 --> 00:27:03.334
But again, students, so you see, like people taking students and student data.

00:27:03.334 --> 00:27:05.356
Again, students, so you see, like people taking students and student data.

00:27:05.376 --> 00:27:07.820
I've rushed to pretty complex concepts already.

00:27:07.820 --> 00:27:11.204
Perhaps I should step back and explain the listener.

00:27:11.204 --> 00:27:24.950
So when you try to model the behavior of people, you can just put a pre-evacuation time distribution on that on a group of agents, right, and just say, okay, on average it takes them from 30 seconds to five minutes to start evacuating.

00:27:24.950 --> 00:27:26.958
Whatever they're doing doesn't matter, it's a delay.

00:27:26.958 --> 00:27:46.794
You can use some decision tree and try to figure out okay, this person needs to identify a queue process, the queue reach a critical level of stress, then inform the person next to them and then start evacuating and assign some values to those to them and then start evacuating and assign some values to those.

00:27:46.794 --> 00:27:53.973
Or perhaps you can observe like 100,000 people and say 37% of population would do this, 15% would do that, and from that have some distribution of tasks.

00:27:53.973 --> 00:28:09.576
What you propose is even further, because you want to observe and then use neural networks or machine learning models in general to provide you that to do this decision making like a human would, which actually could be the way.

00:28:09.637 --> 00:28:35.946
Like you know, the human brain is a machine learning interface, so perhaps this could actually be more natural than we think, to be honest and I mean and if we can boil down what kind of people we're dealing with and again, in a scenario, and especially knowing that certain type of buildings are particularly vulnerable or they host vulnerable populations and have that better characterization.

00:29:06.190 --> 00:29:11.711
I don't know exactly who they are, the people involved in an evacuation, but for instance, there are certain setups when I have people that actually cannot even evacuate and those are the ones that die the most often in fires.

00:29:11.711 --> 00:29:25.002
So, flipping a bit the coin of R-set, A-set into you know, just looking at the tail of the R-set and of course there are, it's not that simple because there are all the crowd dynamics involved and so on.

00:29:25.002 --> 00:29:43.419
But if you look at the tail of the R set and more than the bulk and the first one that go out, basically you can really focus on that part of our evacuation curve and have a better understanding on those, Because those are the ones that drive the R set generally.

00:29:47.029 --> 00:29:47.290
Yeah, havod.

00:29:47.290 --> 00:29:49.454
Now I'm thinking, you know, from, from the building design.

00:29:49.454 --> 00:29:51.439
Where do I put the?

00:29:51.439 --> 00:29:53.544
Where do I put the end of that?

00:29:53.544 --> 00:29:58.056
Do I investigate 99 of the population, 99.9?

00:29:58.056 --> 00:30:01.759
Or do I investigate, you know, one to a million case.

00:30:01.759 --> 00:30:11.598
And the thing is, because it's a tale, it takes you incrementally more to solve for that person.

00:30:11.598 --> 00:30:24.115
I don't want to say problem, but you know you can solve for most of the population represented by your average agent by your ordinary means of escape, that and that will probably work.

00:30:24.115 --> 00:30:29.417
It just doesn't work for the tail of your distribution, the, the population that's at disadvantage.

00:30:29.417 --> 00:30:47.325
But you eventually go to to cases where which is kind of tragic that this is the source of this discrepancy in in the casualties of that population, where it's very, actually very tough to provide for this particular disability or this particular disadvantage.

00:30:47.325 --> 00:30:53.083
And realistically, in the build industry, I don't think we can really solve for all of them, really.

00:30:54.151 --> 00:31:07.865
I mean I understand your concern because it becomes a matter of financial decision making when do you invest and where do you not but I think we are still at the stage in which we don't really even know where we draw that line on.

00:31:07.884 --> 00:31:16.171
So I think the first mission I mean we're not going to be able to solve this for 100, you know universal design in a vacation we are not yet there.

00:31:16.171 --> 00:31:22.933
I mean we are more and more there for accessibility and even there people complain that it's not as good as it could be.

00:31:22.933 --> 00:31:33.058
But I at least trying to understand if we can push the tail or at least characterize what the tail looks like, because at the moment it feels like we don't even know.

00:31:33.058 --> 00:31:41.069
I mean, when you read the codes or you read, like you know, studies that look at the vacation, very often they talk about people with disabilities in general.

00:31:41.150 --> 00:31:44.681
I mean this could be vastly different types of populations.

00:31:44.681 --> 00:31:53.316
It could be someone completely functional because they have created in their daily life a setup that make them functional.

00:31:53.316 --> 00:31:59.398
Or it could be someone that is completely dependent on someone else for daily activities.

00:31:59.398 --> 00:32:01.016
And again, it's not black or white.

00:32:01.016 --> 00:32:02.089
There is a lot of gray scale.

00:32:02.089 --> 00:32:31.971
So the simple thing of trying to understand those shades of gray, let's say, I think we need to understand where we are now, and the feeling is that where we are now is really much ignoring a big portion of that tail, not as small as we think.

00:32:32.090 --> 00:32:34.958
Yeah, I don't want to sound like I'm against it.

00:32:34.958 --> 00:32:57.579
I just know the reality of consultancy and the reality of designing buildings very well and I know that if you go way too far without having a very good reason for going that far, you know, without having a very explicit risk-based proof that this is actually necessary, then eventually you lose it all.

00:32:57.579 --> 00:33:04.655
Like if you go too far, you're going to be replaced by a different consultant who doesn't want anything and they're going to do the job and the building is going to be unsafe.

00:33:04.655 --> 00:33:10.782
And it's not about just, you know, waving a flag and saying we want the best.

00:33:10.782 --> 00:33:16.923
It's about really turning this idea into reality in the most buildings that we can.

00:33:16.923 --> 00:33:18.616
Then it's a success, right.

00:33:19.211 --> 00:33:23.298
And that's why we need two things First of all, to look at solutions that already exist.

00:33:24.031 --> 00:33:27.836
I mean classic case elevators, I mean now for accessibility.

00:33:27.856 --> 00:33:37.942
We have a lot of solutions that rely on this solution, or like alarms Alarms is not that expensive to make, alarms that are more aiming at universal design.

00:33:37.942 --> 00:33:44.703
I mean there are many things that are, I would say, low-hanging fruits towards aiming at the more diverse population.

00:33:44.703 --> 00:34:00.561
But, on the other hand, it's regulatory, because if this doesn't become, you know, if you don't have a push from the regulatory side, as you said, they're going to find someone else that does it for cheaper, to get it approved, and that's one, I mean one of the good things of the EU that they really push.

00:34:00.561 --> 00:34:19.097
The feeling that I got is and also from the reviews and in general, also from the panel when I read the review that they really appreciate the idea that the project has a strong potential for lobbying towards having more, let's say, to striving towards equality and to strive towards change in regulations.

00:34:19.097 --> 00:34:29.755
Because, again, if you think about the accessibility world, what you are saying today, maybe 50, 70 years ago they were saying the same oh, we cannot put ramps everywhere in buildings.

00:34:29.775 --> 00:34:30.717
This is too expensive Today.

00:34:30.717 --> 00:34:31.378
They are right.

00:34:32.061 --> 00:34:35.539
Oh yeah, and today they are oh, we cannot do this, it's too expensive, the building will not have.

00:34:35.539 --> 00:34:48.565
So I'm in the very early phase of this and I'm aware that at some point you will hit a wall because the current regulations are not made to accommodate fully universal design and aggressability for evacuation.

00:34:48.565 --> 00:34:50.596
But I mean somewhere we need to start.

00:34:50.596 --> 00:34:55.257
So at least the idea to have a large research project at the EU level that looks at this.

00:34:55.257 --> 00:34:58.483
I think it's a very good starting point because then you can start quantifying things.

00:34:58.483 --> 00:35:17.016
Okay, if I start saying you have X percent, you know down the line to have a risk-based approach, you have X percent more of people that will not be able to evacuate and you can quantify that, then you have a much stronger argument towards changing regulation.

00:35:17.016 --> 00:35:17.800
Then now we talk very much general.

00:35:17.800 --> 00:35:20.009
Okay, people with disabilities, they need more help and they need more solution.

00:35:20.009 --> 00:35:25.043
But you know we don't have really something to quantify the consequences.

00:35:25.043 --> 00:35:26.710
I mean to that extent.

00:35:27.005 --> 00:35:36.954
I didn't find it in your proposal, but are you also going to quantify the fire site to some extent, Like I could put a question like how big a grossability feature of the building is sprinkler in it?

00:35:37.434 --> 00:35:46.096
I have to be honest, that's not what I put in the application, mostly because, as you know, fire is not the only thing that I'm worried about.

00:35:46.344 --> 00:35:52.014
I mean, it's probably because fire regulations what the drive of the evacuation design, and that's always been also.

00:35:52.014 --> 00:35:53.717
You know, when I was in the interview I was arguing.

00:35:53.717 --> 00:36:02.065
You know, I'm a fire, I'm in a fire safety engineering group and you may think, okay, this guy wants to design a vacation that is also used for whatever earthquakes and things like this.

00:36:02.065 --> 00:36:07.092
But you know, very often the ones that design a vacation is the fire engineers.

00:36:07.092 --> 00:36:12.519
So it's people like me or in my education, the education where I teach the design.

00:36:12.519 --> 00:36:22.014
So I think the, let's say, the A set side of the equation is not something that I will look at in this project, but the main focus would be on the R set side.

00:36:22.014 --> 00:36:41.295
But on the other hand, I think I will lay the grounds for making this kind of comparisons Because, again, if you can then have a counter argument that, okay, I am not accounting for this, this and that, but I could take care of this through, you know, on the ASET side of things, then you will be able to defend your design down the line.

00:36:41.344 --> 00:36:43.293
But I think we are at a much earlier stage.

00:36:43.293 --> 00:36:51.766
We are not really ready to have a full, let's say, health science-based design for a vacation, because we simply, in fire engineering.

00:36:51.766 --> 00:36:52.989
We know very little about this.

00:36:52.989 --> 00:36:58.534
We are not medical doctors and no one really sat down and learned more from this kind of literature.

00:36:58.534 --> 00:37:01.690
It's like one of those things that if you think about it, it makes sense.

00:37:01.690 --> 00:37:09.311
Okay, why no one really go and looked into the accessibility world and the health science world to see how people are characterized when they do a vacation design?

00:37:09.311 --> 00:37:10.413
But no one really did it.

00:37:10.413 --> 00:37:14.157
So I think that's where we have a lot to learn to start the process.

00:37:14.157 --> 00:37:17.047
But again, wojciech, I understand your concern.

00:37:17.047 --> 00:37:21.405
I sit often and you know many of my students end up being fire engineers.

00:37:21.425 --> 00:37:24.614
You talk with them and how do you use vacation models nowadays?

00:37:24.614 --> 00:37:28.108
And they tell you oh, you know, I do one simulation for the worst case scenario.

00:37:28.108 --> 00:37:30.590
And then I ask what is your worst case scenario?

00:37:30.590 --> 00:37:34.556
Oh, that's what the code is telling me to do, and that's we go back to square one.

00:37:34.556 --> 00:37:38.920
We need to have tools to inform codes down the line.

00:37:41.244 --> 00:37:49.010
That's why I appreciate the paradigm shift in the title of the project, because if you really aim for that, that's what we could really use in this field and discussion.

00:37:49.010 --> 00:37:55.061
You just said, Isidar, I had this argument today with the client just earlier this morning.

00:37:55.101 --> 00:37:55.684
It was ridiculous.

00:37:55.684 --> 00:38:00.572
Like, do you want me to design you a safe building or you want number A to be larger than number B?

00:38:00.572 --> 00:38:12.266
Because if you want me to mathematically prove that my ACID is 400 seconds, our ACID is 308, I can tune the model so it shows you the number, not changing a single thing in your building.

00:38:12.266 --> 00:38:16.302
It's not the point to have a number, it's, it's a point to understand the system.

00:38:16.302 --> 00:38:18.452
And anyway, let's go further.

00:38:18.452 --> 00:38:21.744
Tell me how you're gonna approach experiments, because I found it interesting.

00:38:21.744 --> 00:38:26.195
You've already teased the, the vr stuff, but perhaps you can give me a grand image.

00:38:26.195 --> 00:38:27.358
We need more data.

00:38:27.358 --> 00:38:28.688
How are you gonna get the data?

00:38:29.291 --> 00:38:37.146
yes, I mean I have the idea to use several approaches because, you know, there is a lot to learn from different methods for data collection.

00:38:37.146 --> 00:38:52.706
One thing that I mentioned in my application is that I want to do a systematic review of a set of big case studies that we, you know, ground failures when it comes to evacuation and see how much information we can find about the population, and we know.

00:38:52.927 --> 00:38:54.530
You mean, like big fires, world Trade Centers?

00:38:54.751 --> 00:38:56.376
World Trade Center Grenfell.

00:38:56.376 --> 00:39:11.545
I mean, unfortunately, we see I mean when you look at the official reports and also informal, let's say, type of information that very often there is a significant part of the population that has some sort of functional limitation or disability.

00:39:11.545 --> 00:39:23.061
So to try to map out this, because this would be really helpful in understanding what kind of people you know, looking at the ones that have suffered the most, are the ones that then are most affected.

00:39:23.344 --> 00:39:27.135
Sorry, is it constrained to a building or is it more like a community as well?

00:39:28.327 --> 00:39:34.396
I will mostly focus on buildings because, again, there is a lot of discussion about this.

00:39:34.396 --> 00:39:45.789
I discussed this openly also with the ERC panel and the reason is impact, because on buildings you can actually have much easier time with codes and regulations If you want down the line trying to enforce something.

00:39:45.789 --> 00:39:49.733
When you go at community scale it gets much harder.

00:39:50.224 --> 00:39:51.490
But the framework is scalable.

00:39:51.490 --> 00:39:54.052
Let's move forward to how you're going to solve it.

00:39:54.072 --> 00:40:08.860
yeah, and then you know, I will start doing interviews and again I want to really talk with people that have different types of disabilities and functional limitations and understand how they will face such a situation.

00:40:08.860 --> 00:40:28.777
Talk with them how it is, because, as I said, we started I did a pilot like this in Sweden, this project that I had a couple of years ago, in which we started talking with people with different types of disabilities, and you discover a lot of things that are not immediate to you as a fire engineer because you don't experience yourself that type of disability.

00:40:28.777 --> 00:40:52.027
So there are people that will openly tell you, as it is today the building which I'm in I will not even try to evacuate because I won't even be able to reach what I'm supposed to reach or other people that will tell you I have such good experience of moving around, despite I'm dealing with the disabilities, that I have a whole setup for myself to make this work if something like this happens.

00:40:52.047 --> 00:41:01.657
So there is a lot of variety of possible responses, but there is a lot of unknowns as well Because, as I said, there are disabilities where are very little investigated.

00:41:01.657 --> 00:41:31.606
One thing in which we know very little, for instance, is cognitive disabilities that come especially with age, and there my idea is to not because you cannot get informed consent from people that have cognitive disabilities, but I want to talk with caregivers, so the people that are used to work or take care of people that have cognitive disabilities, and understand from them to which extent people are independent and to which extent what kind of task, depending on their experience, people can do during a possible emergency.

00:41:31.606 --> 00:41:43.927
So there will be a part on qualitative data collection and then I will jump into the quantitative stuff and then, as mentioned, I will deploy online vignette experiments, which is a bit what I discussed before.

00:41:43.927 --> 00:42:09.333
Its before like to having some sort of a hypothetical evacuation scenarios, and then you can you know we try to do this in loan, with different levels of immersion, so it will never be like when you are in the lab doing VR, but you can have designs that are not just filling in a questionnaire, so that you can have because you know the problem of questionnaires that people will ask you what would you do, but there are a lot of limitations in terms of validity.

00:42:15.105 --> 00:42:15.746
If I ask you, what would you do?

00:42:15.746 --> 00:42:16.891
But there are a lot of limitations in terms of validity.

00:42:16.891 --> 00:42:19.902
If I ask you, what would you do and you don't have any sort of commitment to the consequences of what you say you will do, and that's where you know.

00:42:19.902 --> 00:42:24.713
We started looking at here and how we can increase validity in the data when you do this type of vignette experiments.

00:42:24.713 --> 00:42:32.688
And then VR, because with VR we can do multisensory experiences and especially when it comes to people with disabilities.

00:42:32.688 --> 00:42:37.025
Nowadays there is a lot you can do audio games, you can have a visual stimuli.

00:42:37.025 --> 00:42:43.130
I just met this morning with daniel nielsen from university of canterbury and we talk about this smell simulator.

00:42:43.130 --> 00:42:46.797
You know they they have a smoke simulator yeah, I, I've smelled that.

00:42:47.706 --> 00:42:48.847
It's really smells like smoke.

00:42:48.847 --> 00:42:50.650
It was ridiculous.

00:42:51.652 --> 00:42:58.320
So I mean or different type of haptic stimuli, and also link this with physiological measurements.

00:43:03.284 --> 00:43:05.414
So try to see how your body responds to these stimuli, and especially with people with disabilities.

00:43:05.434 --> 00:43:16.728
I've been really fascinated by this concept that I read about in the world of psychology, which is called synesthesia, so how you trick your brain into thinking you are getting a given sensory stimuli where you're getting something else, so like.

00:43:17.451 --> 00:43:27.548
The idea is that if you are blind, can you see an emergency sign because you're hearing an alarm, or if you're deaf, can you hear an alarm because you're seeing flashing lights or things like this.

00:43:27.548 --> 00:43:35.815
And in VR you can have very systematic experiments in which you have this kind of stimuli and you can measure what is your physiological response.

00:43:35.815 --> 00:43:48.034
And this has the potential to be groundbreaking, because then you can see how you can actually help, especially with older people that had that kind of sensory information before, that it's declining for them over time.

00:43:48.034 --> 00:43:52.231
So how can I trigger a better response to an evacuation alarm, for instance?

00:43:52.231 --> 00:43:53.407
Are there ways to?

00:43:53.407 --> 00:44:08.284
By using other sensory stimuli, and I know that this is like very lab stage and very, let's say, fundamental basic research at this stage, but I mean that has really a good potential down the line, because to my knowledge, no one did this kind of experiment.

00:44:08.887 --> 00:44:11.114
There were those shaking beds for deaf people.

00:44:11.173 --> 00:44:19.476
I think, yeah, but not in VR, not in VR and not with like, while you are measuring your body.

00:44:20.947 --> 00:44:27.655
So, again, this is again a gap in data that we have and that you have to create to feed your paradigm shift.

00:44:27.655 --> 00:44:31.054
Very, very nice 45 minutes Snap.

00:44:31.054 --> 00:44:34.494
I wanted to talk inclusive machine learning, but I think no.

00:44:34.494 --> 00:44:54.344
I think I'll just congratulate on getting the ERC and now tell me how the hell you got it, because there's a bunch of students listening and a bunch of people who would love to sit where you are now as an ERC grantee, and for each of people who get this prestigious grant, it's been a journey.

00:44:54.344 --> 00:45:02.614
So I would love to ask you about your journey to ERC, to celebrate this as an accomplishment already and what you're going to get in five years.

00:45:02.614 --> 00:45:06.114
I'm going to interview you in five years and we're going to see about that.

00:45:06.114 --> 00:45:07.891
Tell me how you got to this place.

00:45:08.344 --> 00:45:09.108
Maybe I will tell you.

00:45:09.108 --> 00:45:10.313
Oh, I was too ambitious.

00:45:10.313 --> 00:45:12.230
I had to go back to the real world.

00:45:12.471 --> 00:45:13.934
Which actually is fine for ERC.

00:45:13.934 --> 00:45:15.065
That's the beauty of those grants.

00:45:15.065 --> 00:45:17.932
If you said that in five years, that's an information.

00:45:17.932 --> 00:45:19.155
That's an information.

00:45:19.496 --> 00:45:38.371
I think that's what they really like in these projects is that you have a vision and, even if it's a very ambitious vision, that you try to push the boundaries of your field and I mean in my experience, I mean with Consolidator Grant about two years ago I started thinking about this Okay, should I give it a shot?

00:45:38.371 --> 00:45:42.813
I had tried the starting grant, like when I was very young in my career.

00:45:42.813 --> 00:45:44.289
I mean, it was probably.

00:45:44.951 --> 00:45:50.936
When you were starting, yeah, when I was starting and you know you don't even get to an interview and I was feeling a bit like down.

00:45:50.936 --> 00:45:52.951
I would think, oh, I will never have a chance.

00:45:52.951 --> 00:45:54.114
This is so competitive.

00:45:54.114 --> 00:45:56.273
You know the percentage of success rate is so low.

00:45:56.273 --> 00:45:59.916
But then you know, over the years you start growing confidence.

00:45:59.916 --> 00:46:02.675
And then I chat with people that have got this grant.

00:46:02.675 --> 00:46:07.193
I had a chat with Guillermo and we started discussing OK, should I give it a shot?

00:46:07.193 --> 00:46:10.076
And he was giving very encouraging word that he could give it a try.

00:46:10.076 --> 00:46:10.996
You have nothing to lose.

00:46:10.996 --> 00:46:16.240
Of course you have to invest a lot of time in this, but if you don't even go to the interview, generally it's much less stressful.

00:46:16.240 --> 00:46:18.561
I mean, you, you just okay, it's a.

00:46:18.902 --> 00:46:21.110
As all researchers, we all get a lot of rejections.

00:46:21.110 --> 00:46:22.974
I mean, we're not gonna sell the story.

00:46:22.974 --> 00:46:24.206
They were always successful.

00:46:24.206 --> 00:46:26.048
You know we all get a lot of rejection.

00:46:26.048 --> 00:46:30.536
But there I started working because I had this project in Sweden on aggressability.

00:46:30.536 --> 00:46:45.898
There was like a pilot of what I wanted to do and I learned a lot about from the colleagues in the medical faculty, about you know what we don't know, and I started thinking more and more about the idea that I thought this could actually fly at the EU level because the topic is relevant.

00:46:46.125 --> 00:46:59.751
It's the right time because you know of all the big events that we see around us linked to evacuations the scale also sounds like something you could not do with a normal grant, and you cannot really go much bigger than that.

00:46:59.751 --> 00:47:02.099
No, no, I, I, so it's I had no hope to.

00:47:02.181 --> 00:47:04.327
I mean, the type of data collection I want to do.

00:47:04.327 --> 00:47:14.559
I mean, I have no hope to do it with a regular national grant, which you know they are up to 500k, 500,000 euros, something like this that you cannot do this huge data collection.

00:47:14.559 --> 00:47:20.554
You need a team, you need people with different expertise, you need people full-time designing experiments.

00:47:20.554 --> 00:47:22.585
So you know it's very ambitious.

00:47:22.585 --> 00:47:30.235
And then, you know, I started writing, I polished the idea, I got a few people to review it people that I knew got, got this.

00:47:30.235 --> 00:47:43.226
I had a very nice colleague from the transport group here at lund university, got the starting grant and now we have the lab together, right next to each other, and he was giving me encouraging words and we said, okay, I'm gonna give it a shot.

00:47:43.226 --> 00:47:50.376
I wrote the application and then last spring I got to know oh, you get invited for an interview on the first shot.

00:47:50.376 --> 00:47:54.340
So I was like, wow, and now we can talk.

00:47:55.766 --> 00:47:58.853
So I've reached the point where Guillermo told me it's going to be stressful.

00:47:58.853 --> 00:48:00.016
Yes, how good is that?

00:48:00.746 --> 00:48:03.576
And there you know, my life changed, my daily life.

00:48:03.965 --> 00:48:08.137
I invested a huge amount of time to be as prepared.

00:48:08.137 --> 00:48:11.074
I didn't want to have any regret if this will not go through.

00:48:11.074 --> 00:48:13.114
I thought, okay, I will get it.

00:48:13.114 --> 00:48:14.561
I will not get it, it doesn't matter.

00:48:14.561 --> 00:48:17.449
But I don't want to have a regret that I didn't try my best.

00:48:17.449 --> 00:48:24.929
So you know I spent so much time doing the homework that you know all ERC grantees tell you to do so.

00:48:24.929 --> 00:48:27.855
You know I spoke with people that got it in our field.

00:48:27.855 --> 00:48:29.036
I spoke with Guillermo.

00:48:29.036 --> 00:48:31.679
I spoke with Ruben Francesco Restuccia.

00:48:31.679 --> 00:48:36.456
I spoke with colleagues from other fields that I knew from Lund University that got it.

00:48:36.456 --> 00:48:40.411
I started because you don't know the panel, you don't know who will interview you.

00:48:40.664 --> 00:48:41.889
Sorry, what was your panel?

00:48:41.889 --> 00:48:43.432
It was not eight it was social science.

00:48:43.454 --> 00:48:44.215
That's the other thing.

00:48:44.215 --> 00:48:57.786
I strategically chose the panel that could possibly be most interesting in the topic because there is a panel on human mobility in ERC, yeah, which I felt okay evacuation, human mobility, it's fairly linked, but it's social science.

00:48:57.786 --> 00:49:03.617
So I also went into, let's say, as an engineer, to go in a social science panel.

00:49:03.617 --> 00:49:09.929
It's also a risk because you might get people say, oh, an engineer that wants to explain social scientists how to do social science research.

00:49:09.929 --> 00:49:10.269
It's.

00:49:10.269 --> 00:49:38.148
It's big risk because you know there is people with the entire career dedicated, for instance, to qualitative research, to doing interviews, and so I say doing everything to strengthen my profile, having publications in the world closer to the panel, reading up not only the previous panels because those are possibly part of the panels they will interview, but everyone that got the grant in my panel.

00:49:38.148 --> 00:49:39.271
I've read what was the grant about.

00:49:39.271 --> 00:49:40.675
I studied their profiles so I really had a long list of about.

00:49:40.675 --> 00:49:51.311
I had an Excel master sheet with I don't know a hundred names of people that possibly could be in my panel, because they say it could be people from previous panels, people that got the grant or any scientists in related fields.

00:49:51.472 --> 00:49:55.259
Wojciech, I guessed every single person that interviewed me.

00:49:55.259 --> 00:49:58.090
I guessed pure by this study.

00:49:58.090 --> 00:49:59.853
So this is the biggest product I take.

00:49:59.853 --> 00:50:10.916
I studied so much who could be that I guessed every, and I watched YouTube videos how they talked the way, their accent, if there was like an accent that you could not be familiar with.

00:50:10.916 --> 00:50:12.929
What was their core research?

00:50:12.929 --> 00:50:16.487
Of course you know, with 100 plus people you need to do it systematically.

00:50:16.487 --> 00:50:22.027
So I did a number of people per day and you know, when I got in the reviews I saw the that's an overkill.

00:50:23.291 --> 00:50:28.228
I saw the image of the faces and I recognized the faces and I knew each of them.

00:50:28.228 --> 00:50:31.726
What was their, their research, and I had a long list of questions.

00:50:31.726 --> 00:50:34.612
I guess all but one the questions they will ask me.

00:50:34.612 --> 00:50:42.605
So one was coming out of nowhere so and it was someone very tedious into agent-based model why you're not doing agent-based modeling.

00:50:42.605 --> 00:50:44.009
Agent-based modeling is better for this.

00:50:44.009 --> 00:50:50.072
So I probably you know they are bad cop, good cop type of setup yeah and someone was drilling me on this.

00:50:50.653 --> 00:50:56.715
But you know, you know I teach this stuff and I, you know evacuation is my field, so I could really try to defend myself.

00:50:56.715 --> 00:51:00.675
But I felt like now I did everything that I could possibly do.

00:51:00.675 --> 00:51:04.226
Then, if it, because you know it's so competitive that you might get.

00:51:04.226 --> 00:51:09.331
You know, in my, in my panel, I think we were 10 or 11 people that got it something like this.

00:51:09.331 --> 00:51:17.338
So you might get number 12 or 13 out of whatever number they apply, and not getting it and you're still being very good.

00:51:17.338 --> 00:51:21.655
So I say, you know, I go with a clean conscience that I did everything that I could possibly do.

00:51:21.655 --> 00:51:27.518
But then I have also this funny anecdote that they told me from the panel that's what I saw.

00:51:27.518 --> 00:51:29.717
Okay, this is meant to be because the day that I did.

00:51:29.717 --> 00:51:31.306

186 - Egressibility: a paradigm shift in evacuation research with Enrico Ronchi

Listen On

Recent Episodes