Automated Decision-Making in Transport and Mobilities
6 October 2022
Speakers:
Prof Daniel Angus, QUT
Prof Flora Salim, UNSW
Prof Sarah Pink, Monash University
Listen on Anchor
Duration: 24:58

Prof Daniel Angus:
Welcome back to the ADM+S podcast. I’m Dan Angus, Professor of digital communication at QUT and a member of the ARC Centre of Excellence for Automated Decision Making and Society. In today’s episode we’ll be discussing automated decision making in transport and mobilties, and I’m delighted to be joined by Professor Sarah Pink and Professor Flora Salim who are the co-leaders of the transport and mobilities focus area at the Centre.

Sarah and Flora, Welcome. Would you mind starting us off by introducing yourselves.

Prof Sarah Pink:
I’m Sarah Pink, I’m director of the emerging technologies research lab at Monash University. I originally trained as an anthropologist and documentary filmmaker, and at the moment I am one of the Chief Investigators at the Centre of Excellence for Automated Decision-Making and Society, and one of the co-leaders of the transport and mobilities focus area, along with Flora.

Prof Flora Salim:
Hi, I’m Flora Salim. I’m a Professor of computer science and engineering at UNSW Sydney, and I’m also CISCO chair of digital transport. I’ve actually just joined the school about four or five months ago, so it’s been a very exciting and new change. And my research area is in AI for digital transport and spatial temporal big data in time series to understand human behaviour at multiple scale, from individual to city-scale.

Prof Daniel Angus:
Fabulous. When people talk about automated decision-making in transport and mobilities, I’d say the first thing that probably comes to mind is driverless cars, but I know it’s so much more than that, and I’m wondering if whether you can tell us more about future automated mobilities and the opportunities and challenges that this poses to our every day lives.

Prof Flora Salim:
Yeah, so it’s a lot bigger than that because mobilities in the every day life, you get up in the morning, you’re already thinking about where do you want to go today you know, as part of your commuting habits going to work or maybe in the morning you want to send kids to school, and you have that routine. And basically one of the worst thing that can break your day is when your travel pattern is disrupted in the morning, and that actually effects the end user. And as someone who’s actually doing research – data driven – insights into decision making, the transport planners and operators in day to day, that’s one of the things they want to avoid. They want to avoid and actually make sure your journey is as seamless as possible. And how do they actually enable that with data coming from the traffic loop detectors, to smart cars, to sensors on intersections. Not just as seamless as possible, but also as safe and as connected, and also you know if it’s very smooth you won’t be able to recognise there’s any hiccup in your journey. But the problem is these days, there could be floods, strikes, cancellations all the time, and these are the things – how would you recover from these journeys. So, whole journey as a user experience is one thing you have to consider. And moving beyond that is the multi-modal idea of having some possible modes automated and some are not.

Prof Sarah Pink:
Yeah Dan, I think the example of self driving cars is really fascinating and interesting. I think it’s a starting point because self-driving cars, as are electric cars, are a fantastic example of a technology that has been hyped and hyped and has been expected to come on our roads for years and years. And some people say self driving cars are here, they are here on our roads in some places in very very limited ways. But self driving cars, as it was originally imagined and for the purposes it was originally imagined for, maybe never come about. So I think we need to be really careful when we think about future automated mobilities, because what we see are these cycles of hype and hope and then this gradual disappearance. And then the emergence of another technology which is seen as a solution for societal problems. So we’ve been through electric cars – they may be on our roads but maybe not in the way they were originally imagined, self-driving cars – yes I think there will be some self driving cars on our roads in the future, but I think also not in the way it was originally imagined. We were meant to be researching self driving cars in Sweden, going around the road that goes around Guttenberg in 2020. Of course we weren’t, because they weren’t there. But it enables us though, to start to think about other predicted future automated technologies and systems. So, there’s been a lot of hype around mobilities as a service as the next big thing. But where does mobility as a service really exist in a coherent, workable form, in a model now – well no where really, in the way that it’s been anticipated.

But, of course the work that Flora’s doing is so important to this because Flora’s work really enables us to understand why data is important in thinking about future mobility systems and processes. The work that we do in the people program is focussed on understanding what people really want, what people really need. The contingent circumstances they might be in. As Flora’s talked about, your journey might get messed up by some external influence – your journey might get messed up because your kids sick and you can’t take them to day-care or to school, or you’ve got to go somewhere else, you’ve got to backtrack on your journey and do something different. So, we need to listen to both of these sides, as opposed to listened to the hype and the future imaginaries.

The other really exciting and fun thing is flying cars. So, flying cars are imagined to be in our sky’s – piloted flying cars – are imagined to be in our sky’s in 2025. Which are then imagined to then be what’s been quite difficultly called un-manned flying cars. So, there’s also a gendered kind of dimension to this vision. But of course maybe there will be flying cars in our cars, but again, we need to think about the futures of those kinds of vehicles and the kinds of mobilities they entail from the data perspective, and people perspective. Rather from that kind of utopian tech futures vision.

Prof Daniel Angus:
Yeah, that’s really fascinating, and you’re right. Often transport and mobility conjures up these kinds of imaginaries around future cities that may never come to be, and kind of perhaps gloss over the here and now, the reality of the every day experience of users. So, I guess with that, it’s a grwat moment to ask then, the kinds of research. So, if you could perhaps elaborate a little bit on, what are those research questions you’re asking of those systems?

Prof Sarah Pink:
In the people program, and especially in the Monash node of the Centre for Automated Decision Making and Society, we’re looking at some of those diverse experiences of ADM in relation to transport and mobilities. So, for example we’ve been looking at disability and future visions of automated transport and mobility. Jeni Lee has made a series of films working with two blind participants and one deaf participant with whom she’s collaborated, to really demonstrate that we need to start to design from their perspectives, and with them, to really highlight the problems that self driving cars or electric cars, and new mobilities might create for them, and to start to address those questions. So it really helps us to reinforce that argument and to demonstrate to industry partners, to policy makers, and to a wider public, where design needs to start when we account for people and what we really need to think of when we understand what the complexities of every day travel might be. So that’s just one example.

We’re working with a range of different communities here and in Sweden, because also we work with our partner institution Halmstad University in Sweden who have a really important future automated mobilities program of research which accounts for people, especially as well. To try to inject some new vision from the ground into the design department of future automate mobilities systems and technologies.

Prof Daniel Angus:
And what about you Flora, what research questions have got you most interested?

Prof Flora Salim:
There are heaps of course, but just to pick a few. So, I’m also chief investigator in the machines program, and with my colleagues in the machines program, we’re looking at issues of explainability of issues of predictive models. And another aspect is also fairness in the output of a decision-making system. So in terms of transport and mobility. Basically there’s also even part of the agenda and transport agenda from transport New South Wales, or Victorian department of transport, equitable mobility is one of the priorities in the next 5-10 years. And what does it mean? It’s not just – of course a lot of the things we discussed before, like Sarah mentioned accessibility is an important issue – but also thinking about access to transport networks, especially in the time and day like this with things like climate change and the increasing number of floods, and La Nina coming, and you won’t know the next day that you won’t be able to go to work, or somethings disrupted along the way. Tat crisis response actually has become a critical part of modelling in transport and mobility. And the problem that has been around for many years and decades in traditional transport engineering and also transport operation, is the one size fits all policy, where the end user is really removed. So, in our study we’re really very much interested in that fine grain predictive analytics on who are these transport users, what transport modes are actually needed, at what contexts or situations. And if there’s unseen events or unseen disasters, how do the transport networks recover from it? How will people be able to get back to their journey as they planned? How do we even understand peoples intent. So for example in technology like the search engine. If you search, intended human behaviour has been modelled very well. But travelling intent, and that’s actually still a very hard problem, especially now, travel has changed. People can choose when they go to work and when they stay at home. So modelling transport demand with data that is so dynamic is becoming even harder. And secondly there is also new technology, and a push towards 0. For example, electric vehicles, public transport, electric buses – where do you put all these chargers? Where would be the most optimal place to put them in a way that would be a seamless choice for users even as this shift towards electrification, personal car, or buses, but it also doesn’t add a burden to the structure. So that’s one of the things that we look at as well in the machines program. I’m looking at predictive optimised problems, for example. And another example is looking at the third parties. So, mobility as a service that Sarah’s mentioned. One of the minor players these days, is the E-Scooters. So each of the E-scooters company’s has their own recommender system, the app. They can recommend where to get the next scooter for your next journey. Now, each of these is optimised in such a way, only for their scooter, their market share. But each of these is playing at attention in a limited space, in shared lanes, shared with cyclists for example. And sometimes because of that pressure they go to pedestrian footpaths. And that creates an impact, which could be very positive. It could be creating, activating areas that people don’t usually go to. But also can create some problems if they don’t park properly and all this other stuff. We tried to quantify that impact as well.

Prof Daniel Angus:
Yeah, that’s really important obviously. I’m from Brisbane, and everything you’re talking about resonates very strongly. I mean we had our floods recently, and we could see the chaos that that created on the road networks. People couldn’t get to schools to collect their children in the afternoons, and hospitals were completely cut off in some areas because road networks or rail networks were completely shut down in some areas. So, I can absolutely understand why these are really critically important questions. Also that micro mobility of scooters – really important as well – we’ve had some experiments hoisted on us in Brisbane around E-scooters and as you’re saying Sarah, the kind of intersection with accessibility. You know, how do participants who are vision impaired navigate a city where there are these scooters maybe lying around on footpaths, right. So these are really great questions.

I guess I want to pivot now to some of the outputs of what your team has been working on, and I know there’s been recently, a pretty significant scoping report that’s been done. Can you tell us a little about the new knowledge and critical insights you’ve generated in this new report.

Prof Sarah Pink:
Yeah, so the scoping study is one of the fundamental pieces of work that we’ve developed to inform the way ahead for his focus area. And for me one of the many striking findings form the report was actually the narratives about future mobilities and automated mobilities that come from industry, from policy, and from the consultancies, has not changed –  it’s not shifting. The benefits of automated mobilities, self driving cars for example, in particular, have been predicted for a long time to be safety, lowering carbon emissions, and making life more comfortable for users. And we were seeing those in 2014 when I first started work on self driving cars in Sweden. We’re still seeing today in Australia – this scoping was only Australia focussed – and Emma Quilty who did the analysis, and Emma and I did the interviews for another part of the study, Emma did the analysis of the reports and this is what we found came out. And it was quite surprising to me in some ways, and perhaps in other ways it’s not so surprising. We know that these narratives are very embedded and the assumptions and the predictions that they make tend not to shift very easily. So it’s very important for us in our work, to do work that demonstrates that those narratives need to be shifted, that the assumptions that they make need to change so we can actually think about realistic and plausible futures for automated mobilities.

And you know, Flora was saying, how do you use data to predict what’s going to happen, especially when you get these contingent situations, when you don’t know that the weather’s going be. How do you actually work with uncertainty in this area. And how do you work with uncertainty about how people might live with these new technologies and new mobilities in the future.

So, another part of our work has been thinking about how do we think about visions for the road ahead, which Thao Phan is looking at. So, how do we get to understand how people living in cities envisage their own mobilities futures, how do they envisage the gendered aspects of mobility futures. How do they re-envisage them in terms of how they would like to travel in the future in relation to their own situations, their gender, their family situations, and everything else that effects their particular trajectory’s, their commutes. How do they actually feel about that? So, we’ve been trying to work with real people to understand their personal mobility futures, how they envisage them, and what kind of contingent circumstances they might encounter. And that also connects to what Flora was saying about electric vehicles. Yolande Strengers who is an associate investigator in the centre and I – one of our projects we want to connect on together with our findings from our digital energies futures project, with the ADM+S work, we’ve been doing a lot of work there around electric vehicles and electric cars in particular. And understanding how people would like to, and where people would like to charge their electric cars in the future. And our research demonstrates that most people would like to charge their electric car at home in the garage so that they’re in control of it. They can decide when they switch it on, when they switch it off, when it’s going to have a full charge, and most importantly that it’s at home if you need it. Which goes back to the contingency, the idea that people want to feel prepared, they want to be in control of how they’re prepared.

So, when we then start thinking about how do we want to put charging stations in cities, an awful lot of work needs to be done around that in terms of understanding people. And how some of those things might reconfigure in terms of where people could charge, where they would want to charge going forward. So, many questions so that we can get it right. And that’s what’s so important about this focus area in ADM+S, because we have this capacity to look at those questions across the different programs of the Centre. Across People, Machines, Data, and Institutions, and really ask those questions together. Because so much of this work tends to be siloed in term of academic disciplines, but also as Flora was talking about, stakeholders themselves silo the work in their organisations, so we need to get different academic disciplines together, we need to get multiple stakeholders from different fields, we need to do the research and we need to have the conversations.

Prof Daniel Angus:
This is all great. And I think they’re really important point you’re raising here, is the need to think differently and the need to take account of lots of different perspectives in trying to understand the scope of the problems, and I guess indeed to understand the possibilities for the future, to embrace future technologies, but in such a way that they do a service to wider society, and not just reinvent and kind of recast existing and old, out of day ideas around mobility. I guess this is a really exciting moment to talk about the interdisciplinary symposium that you’ve got coming up, future automated mobilities: towards hope, justice and care, on the 20-21 October in RMIT Melbourne, and online, particularly because it’s going to include all these different stakeholders right. We’ve got industry, government organisations, not for profits, academics from all career levels, participating from all over Australia and overseas. Can you possibly provide some more information around that, and what do you hope to achieve from that event?

Prof Sarah Pink:
It’s very important for us to centre questions of care, justice and hope. There’s lots of critical visions of automated mobilities and lots of accounts of all of the terrible things that they’re going the do to our society, the injustices that they’re going to bring about. But more importantly, right now we need to consider what could go right, and the role that we could play in ensuring that that happens. Our symposium brings together those diverse perspectives with panels that bring together those stakeholders from Sweden and Australia. We’ve got panels that show our films, really bring real people and their views and their undertraining into the picture. Flora, you might want to talk about the panels that you’ve been developing as well.

Prof Daniel Angus:
Yeah, Flora, what are you hoping to get out of the Symposium, and what do you think are the most exciting elements of it?

Prof Flora Salim:
So, I’m excited about it and I think it’s great that Sarah envisioned it to be so human-centred, so transport-user first. So thank you to Sarah. I’m given the responsibility to chair one panel session, and in that session the theme is doing good with mobility data. Or mobility data for good. So, mobility data is one of the most privacy intrusive, or sensitive. If someone gets your mobility data, there’s a lot of things that can be revealed about your life. Now the thing is, in fact, we have been touching on and off with smart cars, and we use google maps to plan our journeys so google knows our whereabouts all the time now. So lets say we try to recreate a different conversation about it. Basically lets say we do have the data and the privacy and governance aspects are handled properly in respect to the end users, what can you do with this data? So that is sort of the strand of conversations, and I’m bringing in Professor Daniele Quercia from Nokia Bell labs and also a Professor from Kings College London who’s got a lot of experience in bringing end-user experience when they’re planning their journey- what if I want not just the shortest path, but a happy route on the way to work for example. So that will be a really nice different way of thinking about this. Can you actually command the experience on the end user. There’s also another panellist, Professor Andry Rakotonirainy from QUT CARRS_Q who’s been studying for quite a long time, the human factors in safety in autonomous vehicles. He is also very experienced in actually co-leading the trial of autonomous vehicles in Queensland. And also Mandy Mees from National Transport Commission who’s been looking at personal mobility vehicles, and actually using data to inform the design of personalised mobility vehicles.

So, I think that’s a lot just from that panel, but even with all the other stuff that Sarah’s bringing is, we will generate so much conversation.

Prof Daniel Angus:
I just love the conversation of the happy route. This is something that – you know, I’m a cyclist, I ride from my home to my University every day, and I guess I’ve been doing this. Trying to find the happy route. I don’t want to arrive stressed to my office, and I think that’s a really fascinating concept, and one that I hope will be expanded upon at the symposium, because I love this. I love this concept of hope and justice and care, so I wish you all the best with that symposium, I imagine it’s going to be an amazing success.

Is there anything else you would like to add as we close out the podcast?

Prof Sarah Pink:
At the symposium, we’re going to be celebrating some of our key outputs. Our documentary films, our scoping study reports, launching a new book, and also showcasing some of the really exciting work that our colleagues in Sweden are doing. They’ll be hosting a workshop, and also presenting a very special piece of work which they’ve developed which showcases the whole trajectory of a really interesting piece of work that we’ve been doing in Sweden around the stakeholders from the automotive industry, cities, and academics and designers, to actually create a new methodology for thinking about how to design automated technologies for cities. So we’re very excited to share that work from overseas as well.

Prof Daniel Angus:
That sounds great, and thank you so much for joining me today, Sarah, and Flora.

You’ve been listening to a podcast from the ARC Centre of Excellence of Automated Decision Making and Society. For more information, visit admscentre.org.au

END OF RECORDING