Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Standing Committee on Infrastructure and Communications - 24/09/2015 - Smart information and communications technology in the design and planning of infrastructure

MISKA, Dr Marc Philipp, Senior Research Fellow, School of Civil Engineering and Built Environment, Smart Transport Research Centre, Queensland University of Technology


CHAIR: Welcome. Although the committee does not require you to give evidence under oath, I should advise you that this hearing is a legal proceeding of the parliament and therefore has the same standing as proceedings of the House. I invite you to make a brief opening statement, if you wish, before we proceed to discussion.

Dr Miska : I will keep it very brief. The major concern I have with government and ICT is that government, as far as the transport world at least is concerned and I am working in it, is more or less stuck in the early eighties. You are looking for solutions made in 2015 and nobody is bridging the gap. That is a major concern because a lot of money is misspent by investing in projects that deliver 2015 results and are not utilised at all. Being here in my home state, I know of several projects that are run from within this government that are very encouraging when it comes to research. It is very encouraging to be in the front and at the cutting edge of what should be done. The problem is there is no follow-through internally. While we are doing everything digitally now, there are still several places in the department that are using yellow, pink and blue slips of paper that are being pushed around. That does not fit with later data, it does not fit with all the kind of high-tech ICT. The gap has to be bridged and then government will be able to utilise all the new ICT and the stuff that comes with it.

CHAIR: Thank you.

Mr THISTLETHWAITE: Could you tell us how Jellyfish actually works?

Dr Miska : It is very good that I am speaking after Dr Guy. The base information on Jellyfish is a GIS representation of every transport infrastructure element that is outside. There was this question before about models and how models render themselves useless after a couple of years. That was the starting point of my research. About 10 years ago, I tried to figure out how I could store data that is sustainable over time. As a researcher, obviously every year I come up with a new, great idea that I would like to push out. I had to recreate the underlying data. So we tried a lot of times and in a lot of different countries to put up an umbrella to capture all the information that we could need to be sustainable—and we always failed.

In the last five years, with the support of the local government here, we have established the Jellyfish, which is a GIS representation of every physical object that is outside—a piece of road, a traffic light, a streetlight. If you have it geospatially represented then you know where it is and that will not change—except for a drift of seven centimetres a year. You know where it is and then you can start adding attributes to it. The attributes will become more and more rich as the years go by. A couple of years back, the attributes that we were looking into for our roads were very small; we just wanted to know how many lanes there are and how fast you can drive. Unfortunately, most of the road authorities in Australia and worldwide have no idea what the speed limits of their roads are because they do not know where their road signs are. Again, big data innovations are coming up saying, 'Wow, we have all this information now'—but, again, it is pushed into an abstract model where you put attributes to an ID that is irrelevant in a couple of years. However, if you attach it to a GIS point, a GIS representation, you will always be able to go back to it. It enables you to interrogate data on a spatial level—and that is mostly what we are trying to do. This is a major push moving into the geospatial area. When it comes to most infrastructure, it is still linear based over most of the world. When we look at our road, there is a starting point and there is three kilometres, five kilometres and 10 kilometres away from the starting point; but nobody actually knows where it is. It is very difficult to maintain information, it is very difficult to relate to information and it is very hard to keep an eye on these kinds of things.

With Jellyfish you have the geospatial representation and you build different layers of attributes on top of it. The attribute would say, 'What is it that I actually find there?' It is a traffic light. Okay. To which controller is it connected? You would basically find that there is a controller box somewhere on the side of the road. What is in that controller box? What controller is running on it? We are working a lot on maintenance and transport. And who knows what the cycles of maintenance should be in transport? How much paint is on the road in Queensland or in any other state? Nobody knows. So it is very difficult to think about how you plan and invest. If this huge data is available, you know how many cars are passing a certain part of the road; but you cannot relate it to the information that tells you which piece of road it is, what stage of its life cycle it is at or what the attributes to it are; it is very difficult to overlay all this information.

With Jellyfish we have changed that; because everything is geospatially represented, you can overlay all this information. I agree with Dr Guy as well: all the data is available, but the problem is that there are several people working with the specific data and they are not going to share it. And they are not going to share it because the quality is very poor in most cases. Once you start sharing this information and investing in cleaning this data up, there is no more shaming and naming there.

A couple of years back California opened their public data on car detector counting—on how many cars passed which part of the road. That was established by somebody who was put in charge and came from a completely different world. He came from electricity and he needed to know what is happening in the world where he is trying to operate. In transport, that seems to be not really the case; people tried to hide everything. So there was a big uproar; people were complaining about how much money was being spent on data that was completely useless. Everybody was saying, 'What shall we do? We don't want this to happen.' But it turned out that, after two years, nobody was concerned anymore because suddenly money went to the right places—where the data was actually needed and the money was needed the money was actually spent. In a county where you have 50 cars passing over a year, if the detector is broken you do not necessarily have to update it for the next couple of years because there is no data available. What I am trying to advocate is that, essentially, GIS is the only representation that I have found over my career that is essentially the common truth that is out there and that we can actually measure. The attributes on top of it will change. But if they change over the years, you can just add attributes, and make them more granular if you want, and it will be sustainable for the future.

We should get away from attaching ourselves to applications, to products. If you go from government to government to government, everybody has bought into a certain environment of software packages. All of the data is stored in there and they are locked in. There is no way to break out anymore. When we presented Jellyfish to state government here in Queensland, the idea was, 'The industry will not like this because now you are suddenly telling us we should tell them how we store our data and they should deal with it.' We had a big workshop and we invited members from all the state governments and the industry, and found that the opposite was true. Even the industry people know that their applications and their products are only good for doing a certain subset of requirements that you have. Maybe it would solve 80 per cent, but for 20 per cent they are not ideal.

They are not trying to upsell you on this and say, 'Don't worry about that.' They would rather have you use their tool for the best possible thing you would like to do and to get the best results, or otherwise use something from somebody else. But that was not possible, because you have bought into one environment of applications and could not switch between them. Jellyfish would basically break this. Government should decide. Government should say, 'This is the format of how we store data.' Again, it should be based on GIS. Based on GIS, you have several attributes now that the industry has figured out how to produce software that will actually deliver the tools and the benefits to government and then to use those.

CHAIR: Did that cover the question, Matt?

Mr THISTLETHWAITE: Yes, thanks. Following up, who owns the geospatial data that is collected? Is it the Queensland government or is it QUT?

Dr Miska : It is the Queensland government, but most of the time the data that we are using is being put into the open data portal of the Queensland government a couple of weeks later on.

Ms MARINO: In what you have said, you have focused on Queensland government. We are a federal inquiry here. Given your experience and the comments you have made about Jellyfish and other matters, what are the two priority recommendations that could come out of this inquiry that would have the most impact and be most useful given the points you have made?

Dr Miska : Recommendation 1 would be to standardise how government agencies are storing their data and how they make their data.

Ms MARINO: Federal and state?

Dr Miska : Federal should be the one that decides, and state basically will follow. It is the same way in the grand scheme of things. In my state, I would say that state government goes and the local governments will follow. The push has to come from the top. I have been advocating Jellyfish in Western Australia and we have implementations of it in New South Wales as well, so it is not really just on the state—

Ms MARINO: That is one, what is two?

Dr Miska : The second recommendation is—as Dr Guy said before—to make that data available. If that data is not available, you will have no brilliant mind out there actually trying to help you to utilise that data.

Mr VAN MANEN: Thank you for your submission and your presentation. I have a really interesting question which I did not ask Dr Guy earlier. Here in Australia at the moment we have an enormous infrastructure project being implemented across the country in the NBN. Do you know, as the NBN is being rolled out, if—or to what extent—all of its infrastructure is being geotagged to feed into a GIS model or system so that that is open source so that when projects are rolled out we know where that is?

Dr Miska : If I know if that happens?


Dr Miska : I actually do not know if that happens. I would not be surprised if it is not happening at all. One thing that leads to that as well is that when a lot of infrastructure is put out you have information on 'as planned' but you have no information on 'as built'. If somebody made a last minute change for whatever reason, that would have been approved but nobody would have ever noticed it and it would be on no data set that you are holding.

Mr GILES: The element of your submission which I am interested in you expanding on a bit is this concept of establishing a national data clearing house. I am just wondering if I could draw you out a bit on that.

Dr Miska : Again, that comes from where the push essentially comes from. Being, myself, in a transport field, Austroads is more or less like our agency in that all the state governments together, on a national level, basically make the rules and all of the guidelines. When our work was introduced to them, they were very interested in holding a standardised way of data that would be across all the states and also New Zealand. The benefit I see from the more national standard about this is that there are a lot of states that have money and there are a lot of states that have no money, and to enable the states with less money to get a bit of the benefits of these efforts would be very great. If you have a data clearing house that essentially establishes the baseline of what should be known about networks and there are national performance indicators for transport, for instance, there are still states in Australia—and the whole of New Zealand—who cannot supply the required information, because it costs them months and months of work and they just do not have the budget to do that. If you supply them with a base system that is built with money from the richer states, so to speak, and you have a system that runs nationally rather than in the state itself, you can open it for fairly discounted prices, so to speak, to the states that have less money and get more equity across Australia and New Zealand.

Mr GILES: Perhaps what I am missing—I suspect the fault is mine, not yours—is the clearing house nature of this concept.

Dr Miska : Again, it is almost a question of the governance of data and where data should be hosted. Again, if you leave it to all the different states, everybody will have their own solution. Look at all the data portals that are out from every state now. None of those can interoperate with each other. It is very difficult to pull data from two different states and compare. It is just not possible. If this is run at a national level, you can actually compare things. You can say how New South Wales compares to Victoria, Queensland or Western Australia in terms of what return on investment they get, and you can make a fair judgement. Right now there is no way to compare, because you are comparing apples and oranges and you do not know where it actually comes from.

Mr GILES: Thank you.

Mr PITT: I will declare an interest. I am a QUT graduate, from Gardens Point. I have two things. The first one is that we have spoken a little bit about data accuracy, and obviously we have had a number of submissions about what that level should be. What would you recommend? Clearly the amount of data that will be provided is entirely different at 30-centimetre segments to two millimetres, for example. Do you have some sort of recommendation or comment on that?

Dr Miska : I would say: get it going. Do not be too concerned right now with the accuracy. Right now, the accuracy can be off by 100 metres when you look at the road network. Get started, and the more people use the data the more accurate the data will be. Also, utilise the forces that are already out there. There are so many government agencies that are doing maintenance on infrastructure and looking after things, and all the information that is collected there is not added anywhere. Give them the capability to adjust what is out there. But the thing is always that, if you do not tell somebody where you think it is, nobody can check on it. Tell people where you think it is, and they will adjust it once they are out there. So essentially what we are trying to do here is that RoadTek—which is building the roads, so to speak, for Queensland—would actually go outside to attend to a defect on an intersection but then would be able to check if the data that is available off that intersection for the state government is actually correct. But you have to put something out and improve data over time. I think it is worthless to discuss how you have to go to a certain amount of accuracy before you can start to do something, because it is just too much and it just becomes too big a project. The area where we are very concerned, for us right now, is the Gold Coast area. The improvements that we make in data down there are tremendous, because there is interest right now. There is less interest to find out what is happening in Roma, on that side.

Mr PITT: Unless you live in Roma.

Dr Miska : Unless you live there. But obviously you can focus your efforts in certain areas rather than doing a blanket. I would rather say, 'Stay away from any judgement of how accurate you have to be, get it out there and have the accuracy increasing over time.'

Mr PITT: But clearly, if a standard needs to be set, there needs to be a standard.

Dr Miska : But the standard does not necessarily have to be on the accuracy itself. The standard should really be: how do you store this information? If you decouple your GIS data from all the attributes that you put on top of that, that is the kind of standard that I am talking about. The GIS line might be wrong and you might be a couple of metres off, but that can be adjusted later on, and that will move all the attributes of that along with itself. But it should not be part of the standard. It should be the question of giving somebody a confidence level in the data that you have, and that is something similar to what we do here. Once you know and have accurately checked that the data is correct, you give it a green tick of approval, but otherwise put out what is already out there, and it should not be part of the standard.

Mr PITT: If this committee does make a recommendation, the first thing that the minister or Prime Minister will say to us is, 'How much will it cost?' Is there any way to have any sort of semi-accurate reflection of what the cost would be?

Dr Misk a : I am making a bold statement here, but I think it would cost about 20 per cent of the money that you are already spending on useless data collection that is running around nationwide.

CHAIR: So it is a saving.

Dr Miska : It actually would save a lot of money.

CHAIR: When you are talking about data standards, don't you need some sort of standard if you are going to compare the apples with the oranges?

Dr Misk a : You do. Let us say, for instance, when I look at household travel surveys. As a transport engineer, I need to know where people are moving, so I need this service to be consistent. I need to know that all the people were asked the same questions. That kind of information has to be standardised.

CHAIR: And who sets that standard?

Dr Misk a : Again, there should be a national standard out there, and this is like any national agency that is out there that is working together with all the states and has representatives on them doing something about it. For instance, the Australian Urban Research Infrastructure Network was dealing with a lot of open data that was out there. It was a great initiative—a lot of millions of dollars—and as an academic I am pretty ashamed of the whole outcome of this project, because essentially it is a glorified ABS system. Everybody just went for low-hanging fruit, as it was called. Everybody collected the data they can access anyway, so there is no point to it—not that I have not gotten any money out of it. We have provided datasets as well. But the thing is like, 'So where are we now?' When I went to my partners at state government, they said: 'Why should I get excited? That is information that I already had. What else can happen afterwards?' This integration—that is, overlaying and having attributes from different places that are actually matching—was just not given, because you just put different datasets into one place, and you can now download them from there. but again there is no comparison—nothing on there. So I think that is where the standardisation really comes in. You have to have one top layer: 'This is the information that we store for Australia and hopefully New Zealand as well. This is what there is. It is openly available. You tell us what is happening and judge us and be transparent as a government in terms of why investments are happening.

CHAIR: And then the government would say, 'And measure it to within x centimetres.'

Dr Miska : Correct.

CHAIR: How do you look at those sorts of standards that are flexible enough to respond to technological change? As you said before, it is just constantly changing and getting better and more innovative.

Dr Miska : Correct. Again, that is where we completely agree with Dr Guy: it is a GIS system. The GIS system has not changed in in a very long time, and it is the one place where we have our surveyors out there who make sure that we know actually where our land is, so to speak. With the attribute tables on top of that, as in every good computer age nowadays, you would have a certain type of attributes that is valid for the year 2016-17. There will be an update and you will keep maintaining these attribute sets. You just amend these attributes or add attributes to it, and after a couple of years you would retire the old datasets, because you would have requested that all the different states, and possibly all the local governments, have updated their data in a five-year cycle.

CHAIR: Anything further you want to add at this point? Gosh, we are ahead of schedule. It is unusual for us. Thank you for attending the public meeting today. The secretariat will send you a draft transcript of proceedings so requests can be made to correct any errors of transcription. It would be helpful if you could send any additional information—or, as I suggested before to Dr Guy, if you want to respond to evidence from other witnesses, that would be wonderful. Thank you very much.

Proceedings suspended from 13:58 to 14:15