This week Louise and Gráinne catch up with Dr Suzanne Little who talks about her research into automated vehicles and the problems of unpredictable people; Long Pham discusses her journalism career and how it relates to her current role, while Professor Brian Caulfied discusses Covid's impact on his site at UCD and which Insight member he would bring to a desert island.
This week Louise and Gráinne catch up with Dr Suzanne Little who talks about her research into automated vehicles and the problems of unpredictable people; Long Pham discusses her journalism career and how it relates to her current role, while Professor Brian Caulfied discusses Covid's impact on his site at UCD and which Insight member he would bring to a desert island.
Grainne Faller 0:04
Hi, and welcome to episode four of The Insight Podcast. Please rate review and subscribe wherever you get your podcasts. How are you doing Louise?
Louise Holden 0:12
I'm doing great. Thanks Gráinne, looking forward to the show. Later on, we're going to be talking to Professor Brian Caulfield in UCD.
Grainne Faller 0:18
Yes, we're also going to be talking to Long Pham who's AI for EU Community Manager down in UCC. I had spotted her bio on the website when we were working on the website, and I thought she looked like an interesting person to talk to. Turns out I was right, she had a very interesting story to tell.
Louise Holden 0:35
Very good can't wait to hear it. But first it's over to DCU and Dr. Suzanne Little, she's a Principal Investigator with Insight, and she joins us to talk about her research.
Suzanne Little 0:43
Right. So within Insight, I mostly work with multimedia analysis. So my my background, my interest is really in how you get information out of images and video, and the associated challenges that we have with managing the volume of data that you tend to get from those sorts of applications, how you index it so that you can find and understand patterns quickly. And the wonderful thing about this is it's got so many possible applications. So I've done research work in technology enhanced education, in health care, in security, in autonomous vehicles, in sport - there's always something. And I have the coolest demos because I work with video.
Unknown Speaker 1:26
Is there anything you're working on at the moment that you'd like to talk to our listeners about? Or is there a completed project that you think might capture the imagination in our 10 minutes?
Suzanne Little 1:35
Yes. So recently, we've had a few projects around instrumented vehicles. So so these are the self driving cars that everybody everybody wants to have. And most instrumented vehicles, even even today's vehicles where things are on the road for safety systems, they have multiple cameras. So it can be six to eight cameras looking 360 degree view around the car, and lots of different sensors, so LIDAR and radar, various various things like that. And I've had a couple of projects recently, and I've got a few current students who are working on how to analyze the video coming in to detect and track and interestingly predict the trajectory of objects around the car. So pedestrians and cyclists and vehicles. Obviously, this is this is really interesting to work on. There's there's lots of people who are very keen to see this sort of technology in use. And it it poses lots of very interesting challenges for us from that video analytics perspective. So being able to work in all sorts of situations when it's dark, or it's raining, or you're driving on a country Irish road, versus a road in India or Mexico or Australia. So that's, that's one that's a lot of lot of fun to work on. And also is a really useful potential contribution.
Unknown Speaker 2:52
In that sort of, especially with the predictive stuff, do you need to work with a lot of researchers from different fields? I imagine the possibilities when it comes to actually predicting behaviors on the road - I mean, there could be, you know, sort of driving cultural elements ,there ethical elements, you know, do you, you know, err on the side of absolute safety? Or do you have to kind of play with that a little bit? Is it at that stage yet? Where all these other disciplines being brought in? Or is it just still looking at the technology?
Unknown Speaker 3:22
Absolutely, yeah, no, it's definitely it's definitely a combined approach from lots of different partners. So one of the projects I have done recently was a European Union funded project under the Horizon 2020 framework called VIDAS. That finished last year, and that had partners from many of those backgrounds that you mentioned. So there were partners there who were specifically responsible for considering the ethical issues. There were others that were looking at the user interface issues, how do you work with the human driver, and then we had people with experience in things like law and the insurance industry, because all of these have have factors in and a part of it. So it's absolutely a big part of it. And it's interesting that you bring up the the cultural challenges around driving, we had some really interesting conversations about the difference in driver attitudes between, say, Ireland, and then I'm from Australia. So we have slightly different situations we have to deal with kangaroos. And then other people on my team come from India, or from Mexico or from China. And they would look at the kind of sample data and scenarios and think, well, this isn't really what we'd see in our country. So one of the advantages I think we bring as Insight with our variety of experiences to projects like this, is those different perspectives and those kinds of questions. We don't have answers to all of those questions yet. That's That's why this is still an active research area.
Unknown Speaker 4:41
It strikes me that a lot of the stuff that you're working on, it's incredibly new. I mean, when you started out was the stuff even on the radar? It must be kind of how I guess what I'm asking is. Can you tell us a little bit about what brought you to this?
Unknown Speaker 4:57
Yeah, that's a good that's a good question. And you're absolutely right. What we're seeing at the moment is the duration of a PhD is four years, most most of them. And what we're seeing now is people are getting to the end of the four years, and in that time, there's been such a paradigm shift or a thinking shift in the sorts of technologies they use that we're having to say to them, well look, you know, when you're when you're asked, you know, why didn't you use this technique, your response is, 'because that technique didn't exist until six months ago.' But now it's technique that everyone's everyone's using. So we've seen that several times. So that's exciting. I mean, it's challenging at times. But for myself, I guess underlying it all has always been this interest in what you can do with images and video. And then obviously, as the the techniques have changed, or as the particular availability of data. So that's, that's really been the biggest shift. Now, it's still something that occasionally we struggle with to get the type and the quality of data that we want. But the volume and variety of data sets that we can use to work with and to train on, that's been a massive change. And for the better, like it's really helped progression of research in this area
Unknown Speaker 6:06
How comfortable would you be behind the wheel of a self driving car yourself at the moment?
Unknown Speaker 6:13
Probably not that comfortable. And not so much because of the limitations of the technology, because the technology is kind of amazing. And it's not perfect, which is why you you have reservations. But it's it's other people. So the current technology, I think would be would be really good in certain circumstances. So controlled situation. So this could be things like set highways that are only intended for your autonomously controlled vehicles, or, you know, environments like warehouses, or docks, or campuses, even, where we know that. But there's several illustrations of how either the unpredictable behavior of pedestrians, who will do things that you don't expect them to do, or the inability of the cars to perhaps reason, in an unseen situation, there's still there's still a lot of development work that's happening on on that. So I don't know that I would, right this moment. But I think the technology, even at last three or four years has shifted so much that it's it's not out of the question anymore.
Unknown Speaker 7:15
How have you found your research, your work and everything, how have you found things with the pandemic happening lockdown? Are you affected? What's what's going on?
Unknown Speaker 7:24
It's, it's a tricky one. So yeah, in a certain sense, we are we are affected. And we're affected in ways that maybe at the beginning, we didn't think. Now, realistically, I mean, I work on a computer, so I am incredibly privileged in that I can continue my work with very little real change, you know, it's it's, uh, occasionally slightly more frustrating. Everyone gets very sick of zoom, even people that work with video, but you know, compared to many, many people, it's not something that that I think, has been been a massive problem. The kind of the unforeseen issues are more around the the social aspects or the or the collegial aspect of work. It's amazing how much you don't realize until it's gone the the incidental corridor conversations and how big of an influence they can have. So the the quick chat with a student as you pass them in the kitchen, where you solve their problem, or where you find out that there's a problem. Now, every communication has to be deliberate. And that's a barrier. I think that's that's been the biggest challenge. And this, we're trying different things. So we've got, you know, chat channels, and we've got social activities, and we're working on this. But I think that's that's the unexpected difficulty. The research. Yeah, there's been things about getting remote access to systems, but we're computer scientists, we can figure these things out. The collaborative work. That's the that's the challenge.
Unknown Speaker 8:50
I did have a question in relation to some of your other research related to stadiums and smart stadiums. An awful lot of that research is obviously about capturing data that's generated by crowds. And we wonder, will we ever see a crowd ever again? Is this going to be an issue if it keeps going? You'll have nothing to research?
Unknown Speaker 9:09
Well, I would imagine that yeah, there's definitely going to be be changes in how we perhaps interact for a long period of time. But to be honest, actually, some of our research and technology will will contribute to being able to better manage that. So I know that Leo Gualano and Jiang Zhou in High Data, which is a company that they're working on as part of part of coming from their their research work in Insight. They have been looking at how they can use their edge processing video systems to look at observing and measuring occupancy of areas. And this is really important because the video is processed on their device. It's never transmitted or stored. So you never have to worry about actually having video of people and their faces which which presents a privacy issue. Instead, what you get is this kind of dashboard, this report showing you sort of the level of activity that's happening in an area. And the potential for that is maybe to observe something like a campus over a long period of time. And you're you're sort of measuring and monitoring the overall adherence possible. You're spotting bottlenecks in the way that signage has been set up that tend to encourage people to be in a in a closer arrangement than you might want them to be. So while you know, perhaps we're not going to see the eighty two and a half thousand odd in Croke Park that we used to have for some of our research. This really brings new research challenges and new questions that we are well positioned to to try and contribute to providing solutions for
Unknown Speaker 10:50
Long Pham is AI for you Community Manager down in Insight@UCC. She's had a long and varied career spanning the globe working for IBM as well as other research centers, and she has a research career herself. Prior to that she was a career journalist career that culminated in becoming managing editor of the Vietnam News, a daily English language newspaper in her native Vietnam. Long, you're very welcome, Louise and I thought we were the only journalists in Insight but it turns out not to be the case.
Long Pham 11:18
That's true. I had almost 10 years of a journalism career. I grew up from reporter to columnist to senior editor and to managing editor when I resigned from my news organization.
Unknown Speaker 11:34
And was this mainly print based or did you work in other mediums?
Long Pham 11:38
Print based. It's is a national English language daily newspaper, back home in Vietnam,
Unknown Speaker 11:44
And would any stories in particular stick out from your 10 years there?
Long Pham 11:49
So in the in the environment that we are in right now with the global pandemic. I am very much recalling my years and days that I was covering the SARS outbreak in Hanoi where I was working for the news organization. Actually, I was listening to the radio podcast the other day, when a reporter was talking about coming in and out of the ICU across the country to really record and bring the real situation and analysis of the pandemic to audiences. And that recalled my career as also someone who went into the ICUs, and we was very much a little bit afraid and now many friends would say that I was crazy at that time. But because the quality of the coverage and also the need to bring the whole pictures of what it would be like in an ICU with people who have contracted that disease at that time it was SARS outbreak. So then it was the responsibility to the people, bringing them the reliable information and balance coverage so that they would decide for themselves what would be the wise actions for them to take. So that's what it was for me during those years. And I won the National prize for that year of coverage.
Grainne Faller 13:21
Really?
Long Pham 13:22
Yeah,
Unknown Speaker 13:23
That must have been an incredibly scary time. Because the one thing we say about COVID well is at least it's not SARS, but I mean, the death rate was so high, it was such a scary, intense disease. This is too, but it is it's not on the same level, I don't think at all.
Long Pham 13:41
In Vietnam, we are a tropical country. So we have kind of different pandemics all the time you know, it's probably not the global one. We have the epidemic mostly every single year. So somehow people were already kind of having the idea in the back of their head of how they could cope. But like SARS at that time was a different kind of level of intensity. And we had WHO experts there with us we had CDC people there with us from different countries. And we were trying very hard to try to contain it. We were very lucky that it was finally contained after I think it was six weeks of a very intense battle. All different forces joined together, then we we had it under control. But the thing is that as a journalist as I was, the responsibility of getting people informed and getting people that reliable information as you see with the infodemic right now that we are facing, there is a lot of that going on and AI has been somehow helping that's why it is related to the works that I am doing right now with the AI4EU project is actually that how we are making people informed based on the reliable information, reliable analysis. And I think that is the golden thread throughout my career is actually that - trying to enable people, trying to help people to get the quality information that they need. And I believe that people are smart, they can make decisions for themselves, they know as long as they are being informed. And that is also something that AI in Europe has been striving to talk to is actually trusted AI. And last week, the European Commission was talking about building an ecosystem of trust and ecosystem excellence for AI in Europe. And that based on the trustworthy AI, and human human centric AI, which is I think that even though that I am not a computer scientist, or in any of those mainstream careers in AI, but from the communication point of view, and also from understanding the the nuance between science and communication, I think that my role there is continuously like, what I was. Trying to bring people's researchers, businesses with the reliable information in AI, helping them to Firstly, empower them, also enable them, and also trying to get them to work together.
Grainne Faller 16:52
It's funny, when I was looking at your, you know, your, your bio, on the on the website, I was wondering how you draw a line from the beginning of your career to where you are now. But you've explained exactly, that there really is a connecting thread there, and has that been delivered, a delivers all the way through, because you have done a lot of different things.
Long Pham 17:12
I think it has been that way, since I started with my journalism career in the business that I'm in now, which is trying to communicate between ordinary people with the science and vice versa. But again, we see the the growth of not so reliable information, a lot of conspiracy out there, and all that kind of thing. And then the the attention spans of people is getting kind of shorter and shorter. So then you also have to, to evolve. And you also have to learn how to try to communicate with them in the ways that are effective, bringing them the reliable and trustworthy information, especially when it comes to AI when people would have a lot of curiosity at the same time. But it's something that would kind of scare them because of all the possibilities that jobs could be replaced by AI, by machines, by robots, and all that kind of thing. So how you try to see from their point of view, and then you understand also from the scientists point of view. So it's you trying to kind of get the message there, in the ways that people can understand. I learned a lot since I take on this, this opportunity. The skills, journalists help a lot with the content generation, it has been very helpful. But still, you got to learn, you got to continuously see opportunities that you can help more people. And overall, you can help people to unite. If you know, where would be the winning point for everyone.
Louise Holden 19:08
We're going to talk now to Brian Caulfield, who has been working like everybody else under the restrictions of lockdown. Let's find out how it's been going for him.
Brian Caulfield 19:16
There's been some positives, and there's been some negatives and probably more more negatives than positives. But on the positive side, I think because we have all been fixed and rooted to the same spot, we've had a little bit more bandwidth to get together for discussions and around collaborations and plans. Albeit virtually, getting together virtually is not half as good as doing it face to face. But we've I think we've probably had more opportunities to get together than we would normally have had. So that's on the positive side on the negative side. It's been very challenging to keep it going. We had a number of months. I mean, a lot a lot of the work that we do and incise we we have to do Face to face research. I remember one of the meetings early on with the Vice President for Research and Innovation in UCD. And a comment was made is that all in inside you guys probably okay, because you know, an awful lot of the work that you do there is based with working with data, which is a very logical assumption to make. But an awful lot of people in inside have to generate data as well. And those people in Insight are probably in the minority. But face to face research was was completely off the agenda for a long time. So you can have as many Zoom sessions as you like planning research and planning collaboration, but the rubber eventually has to hit the road, and you have to put sensors on people and capture data. And we had to put an awful lot of energy into getting back to face to face research. So now we have a new, we have a new model in place that enables us to do face to face research again, but it's not like it was. And it's not quite as obvious that people will be willing to volunteer for face to face research like they would have done in the past. yet. So those those types of things have been big challenges, being able to go out and speak to companies about potential collaborations have been big challenges. And the decision making process and potential industry collaborators, as has been very much stymied. And our ability to get out and build collaborations around European projects, or even further afield has been has been quite difficult. So but you know, in the grand scheme of things, these are not, these are not big problems,
Unknown Speaker 21:35
Can you just for people listening who might not be familiar and describe quite you doing incise at UCD, and where you sort of sit within the Insight ecosystem?
Brian Caulfield 21:46
I guess, I'm in the Sensing and Actuation group, we're interested in what we can achieve with data primarily from wearable sensors. And the data we're primarily talking about is data relating to human movement and behavior. So we capture at a macro level people's gross behaviors, how often they're moving, how many steps they're taking per day, how many hours they're sleeping per night, you know, so we love applications, where we're looking at that macro level of behavior, how people construct their behavior throughout the day, to micro level, where we might be looking at their biomechanical form as they, as they perform a specific task, or some physiological measures as they perform a specific task. And we're interested in in a variety of application contexts, trying to use that data to better understand human behavior and human performance. And then ultimately, try and close the loop by moving from that understanding to actionable knowledge that can help people have better behaviors or better performance. And the type of application settings we do that work in is is in you know, sometimes it's a wellness applications, sometimes it's a sports, whether it be a leisure or recreational sport, and a lot of the time to it's in health care.
Unknown Speaker 23:05
We put out the call Brian, across Insight, for questions for you. We got quite a good response. So if you don't mind, we might ask you a couple of those and anonymously. Some of them are quite sensible. For example, what are the three things that are most vital to the success of insight two, in your opinion?
Brian Caulfield 23:27
Three things? Well, ultimately, Insight's about people. And the success of Insight is that that's the most important thing is that, you know, if we have a phenomenally talented bunch of students and postdocs working in Insight, who make the investigators look good, being able to consistently attract talent into Insight is is, is our biggest determinant of success. Outside of that the investigators doing the hard yards of making sure that we keep the, you know, at a at a planning level, and at a level of directing the centre and at the level of making sure that the funding taps stay turned on, I think is important as well. I I'll have to think about a third one. I'm not quite sure. But I mean, I'd be what like I kind of think that I'd nearly lump kind of the investigators and all of the administration and operations, part of Insight kind of like our job is to is to provide a bit of direction and keep the show on the road. And it's the it's the talent that that realizes the outputs. Two out of three is enough.
Grainne Faller 24:42
Fair enough. We have touched on this, but have you seen about how COVID-19 has impacted the Irish research ecosystem in general, and the past few months?
Brian Caulfield 24:51
Yeah, the uncertainty around everything is, I mean, it's just thrown a massive rock into the pool that, it's disrupted everything. And some of the things that has sparked off have been very positive, you know, if you look, take, take on the positive side, look at the reaction of the research community to the COVID response or to the COVID pandemic, you know, there has been a massive mobilization of effort to get the best out of the expertise and an infrastructure that is there in the research ecosystem. And Insight has been really, really, really strong in its response, which is fantastic to see. You know, coming from my research background and my application domain, I haven't had much opportunity to directly apply myself, but I see some colleagues in Insight doing amazing work in indirectly, you know, assisting the government efforts in Ireland and beyond. And there's other really interesting research and insights, you know, emerging, Tomás Ward and myself are just about to launch an app, an app based study where we'll be collecting wearable sensor data from people and trying to see whether or not that can provide an early indicator of a Covid infection in people, Madeleine Lowery and Emer Doheny have a fantastic project just kicking off at the moment where they're using mobile phone to record breath sounds, and then processing the audio signals, again, to try and aid in the diagnosis and management of COVID-19. So there's some really interesting pockets of research emerging within Insight. But that has been seen throughout the ecosystem. And that's on the positive side. But on the negative side, I think our ability to go out and disseminate our work and the ways that we used to, has been phenomenally impacted. And it's very difficult for us to to get out there and to properly interact with the wider international scientific community. And people can cynically look on from the sidelines and say that, or they, you know, researchers love going off to conferences, or big junkets and it's very enjoyable, sometimes going off to conferences when they happen to be in nice places. But you do, certainly your research does benefit an awful lot from the engagement that you get with the wider research community at those conferences. And that's, that's just been pulled from us. And you absolutely, you can't even come close to replicating that with digital platforms, the ability to interact with industry, it has has been severely impacted. Because quite justifiably, an awful lot of R&D programs in industry, particularly where they were more long term discovery type of research programs that have been undertaken by industry and the type of work where they like to collaborate with academia have been put on hold in a lot of cases. So it's difficult for us to build on existing collaborations, or to engage in new collaborations with industry. And we could see the effects of that for a while.
Grainne Faller 27:53
We're almost out of time. But Brian, a really important question came through from someone who I would say should probably remain unnamed. And
Brian Caulfield 28:04
Oh, good. (laughs)
Unknown Speaker 28:07
But it is important to I think Insighters have a right to know, if Brian was marooned on a desert island, Who does he think would be the most useful member of the Insight team to have with him and why?
Unknown Speaker 28:17
Ah it would have to be Donnacha O'Driscoll. I don't know if Donnacha asked that question. But it would have to be Donnacha. Apart from the fact that we could probably talk about running most of the time. And I'm sorry, Donnacha. But if we were to be marooned on a desert island, and we were we were we were going to have a competition, I'd be the island champion in running. Because I might not be brilliant, but I know I can I know I can beat Donnacha in a race. Yeah, I mean, he, you know, I think Insight in UCD and maybe in other parts of the ecosystem as well would grind to a halt if Donnacha wasn't there. So I'm sorry, everybody else who was expected to be named. You all come a very distant second.
Unknown Speaker 29:11
I think that's very fair. We'll have to have Donnacha on and see who he chooses. I think that would be an interesting one.
Brian Caulfield 29:20
I'd be shocked if he chose me. He gives me solutions. I give him problems. Yeah,
Grainne Faller 29:30
I think we all have that relationship with Donnacha.
Unknown Speaker 29:36
Thanks again for joining us and to all our listeners over the last few weeks, the Insight podcast wouldn't be the same without you. If you have any ideas for interviews, if you'd like to come on the podcast, we'd love to hear from you. You'll get us at info@fhmediaconsulting.com. We'll be back next week with more great interviews. Hope you can rate, review and share and that you'll be with us next Wednesday. This has been a snoring dog production on behalf of the Insight SFI Research Centre for Data Analytics
Transcribed by https://otter.ai