Dr Ian Ferguson, Lecturer in Computer and Information Science demonstrates that Digimap is not just for geographers.

Ian Ferguson, living proof that Digimap is not just for geographers. It feels a bit like an alcoholic anonymous meeting: ‘I don’t have a geography O-level’. The reason for that was a dreadful teacher but I’m sure you’ve all had that experience I won’t bore you about it now. First of all my thanks to the organizers for inviting me to come along and talk about this, as you can see I’m actually from a computer information sciences department at the University of Strathclyde- the West Coast mafia are out in force today. So what am I doing here talking to a bunch of geographers about GIS and Digimap data? Well, it’s a bit difficult in that when I was first asked to do this one of my criteria was ‘don’t come and talk about anything technical. They’re a bunch of geographers don’t come and talk about computer science’. So I went back to my department and said ‘Can you give me some advice? What would you expect from a bunch of geographers?’ and they said, ‘Well, pipe smokers, probably a lot of beards – and you should see the men!’

So anyway, I have tried to put together a presentation about a project that one of my final year students did. My research background is from the software systems group – the delightfully named SoSyG. And my particular area of interest- context sensitive computing, mobile computing, and location aware computing. Now we’ve already had a reference to the basic principles behind this earlier on – the martini principle: anytime, any place, anywhere is - where we’re interested in making data available, via mobile devices, mobile phones, iphones, laptops- you name it we’ll have a go at it. Last year one of my project students Liam McBrian – more of him in a moment – produced something called the ‘air safe project’ which is an example of this kind of computing, the reason is interesting of course in that uses Digimap data.

Before I talk about that I hope you will indulge me for one moment. One of these strange coincidences that happen in life occurred to me – I was talking to an old uncle of mine over the weekend. I said I was about to come to Edinburgh to give a talk about mapping and digital data – he’s a graduate of the University of Edinburgh geography department circa 1950 kind of thing so I thought I’d tell him about it. He said “Oh, oh yes, your Uncle Walter will be very proud of you doing that”. I said “Uncle Walter? Never heard of Uncle Walter? Who’s Uncle Walter?” “Oh he had something to do with maps – Walter Smith” “Who was he?” “Oh, he was the director general of the Ordnance Survey”. I’d never heard of Uncle Walter before this time – anybody know anything about him or the history then please do tell me about it, it would be fascinating to know about it.

Anyway, I feel we must do some work, enough of my personal bits and pieces. The air safe project – final year project, when we sent our final year computer science students something – (struggles to turn on the screen) we’d preferably give them some technology that works – and we asked them to develop a significant piece of software to do ‘x’ where ‘x’ is hopefully something interesting. So I had Liam who is now MEng computer science – his chosen problem domain was the area of general aviation or light aircrafts such as the beastie that you see sitting up there. He wished to address a particular problem- if you’re a nervous traveller, if any of you flew up to Edinburgh you might want to miss the next few slides. Here we have a view from inside a light aircraft - that’s the cowling, the nose. And there it is heading over the country side – and, oh dear – we have some cloud that it is about to head into and this is not good, this is bad news. 30% of all aviation – this isn’t just light aviation this is all aviation fatalities are to do with controlled flight into terrain, where the aircraft was perfectly air worthy, the crew were of sound body and mind, they just didn’t know where they were and failed to miss the mountain. I could talk about the technicalities of instrument flight rules and visual flight rules but we’ll give that a miss this morning I think. So as you can see the consequences of getting this kind of thing wrong are fairly bad. So, the solution that we thought we’d give a try to was something like this: Air Safe - the sat nav of the airways. An attempt to save lives by the application of computer graphics, Digimap and GPS. And in particular what we wanted to do was to come up with something can we use a GPS position and a digital elevation model in real time, in the aircraft to give the pilot a warning that they’re about to have a bad day – that something very nasty is about to happen. In the background you see the final solution, we didn’t get quite there in one step. What we thought of doing was replication the moaning Minnie that you’re all familiar with from your sat nav in your car. Why didn’t we go down that route – well, you’ve all experienced moaning Minnie the sat nav in your car I’m sure and that’s one of the first reasons. But, one of the principles of mobile computing -so excuse me for a minute if we do talk about mobile computing rather than Digimap – is that the user’s attention in situations is by far the most precious resource. Yes, we’re all worried about disk space, about memory, about power, about battery life, but the most precious resource is the user’s attention. So HCI, Human Computer Interface, was very much the key to this problem – how we presented this marvellous Digimap data.

Now there have been around for a while things called synthetic vision systems. Basically that’s what you need to know – they show you on screen what you would see out the cockpit if it wasn’t covered in cloud. So here’s a nice professional one from an air bus somewhere, and sure enough we can see in the background there are some hills which probably wouldn’t be a good idea to fit. So the technology in a sense isn’t terribly new – but – reasons why we wanted to do it I will explain in a couple of slides time, but this is roughly what we ended up doing. Something like this- Here’s our airplane heading towards what is affectionately known in general aviation circles as cumulo-granitus – a mountain covered in cloud. However, he has on board our system, it is receiving a GPS signal from the satellites, so it knows where it is. We also have on board a digital elevation model and I’ll let you guess where we got that from. Actually we don’t take the digital elevation model around with us, we pre-process it quite significantly for reasons I’ll come onto. And what you seen in the screen is the pilot has his laptop or whatever in front of him and you see a picture of what you would see out the front if the cloud wasn’t there. If you really want to hammer home the point put a nice big cross on it which says if you don’t do something this is where you’re going to end up very soon. So it works something like that. The challenges for a computer scientist were actually doing this on the cheap as it were. We’ve got the big professional systems, they’re fairly common place, you’re talking about minimum 10 thousand pounds I would have thought, maybe 10 thousand dollars in the states. People don’t buy them to put them in light aircraft, that’s probably more than some of the light aircraft are worth believe it or not. All we wanted to do was see if we could take a cheap PC – and I could go out of here and walk 500 yards down the road and I bet I could buy a PC for under £200 these days. So the challenges- speed, can we do it fast enough? Can we get the data from the database onto the screen quick enough with our system. Can we store all of this data? Have we got sufficient capacity to take it all with us in something light enough to take in a light aircraft? Where are we with mobile devices these days, are they good enough? Can we actually fit a display onto the instrument panel? And possibly most important – lastly, usability – can we present the information in some kind of fashion that the pilot really doesn’t have to apply any effort at all to interpreting it. And that very much was what the key was about.

So, ok, we’ll have a go with a DEM where are we going to get it? Well we could use the one from google earth, freely available. Don’t know quite whether it’s legal to use that with the licensing requirements? We could actually pinch the data from Microsoft flight simulator – a well known game. It’s all in there, there is a version of the terrain data in there – is it accurate enough? Probably. Is it easy to get? Yes. Is it legal to use it? No – there’s that one out. NASA - the original survey - no sorry don’t know anybody at NASA. The Ordnance Survey – bet they’ve got lots of data – and around about that stage one of us struck gold and we heard of Digimap and say “Ahh, this looks promising!” It’s free to academics – the magic word free came to mind. Thought – ‘Ok, let’s see if we can do it with that’. I should at this point bring out one of my visual aids here that I’ve actually been playing around with this kind of data for a long, long time. The reason I got into it ages ago was that product from the Ordnance Survey which some of you may or may not recognize. It basically allowed you to just put some maps on your PC. That particular version does have the digital elevation model data in there in a completely proprietary form which you can’t access or do anything useful with – unless you spend rather a long time reverse engineering it – about which we shall quietly draw veil over proceedings.

Ok, so the first prototype looked something a little bit like that. This is the West Coast of Scotland, there are some islands that look a bit like that. Visually perhaps a bit unexciting- but we’ve got sea, we’ve got mountains, we can see roughly where we are. More or less one of us kind of said wouldn’t it be nice to know which mountain you’re about to crash into at that stage, and we thought ‘can we stick some labels on these things? Can we name the islands etc.?’ Oh right ok, we’re really pushing it on these small computers to be able to display this stuff in real time anyway. If we start trying to texture the stuff as it’s known in computer graphics, if we start trying to lay the map on top of the terrain, can we actually do it? And the answer is obviously yes, otherwise I wouldn’t be standing here, but we ended up pre processing a lot of this stuff - turning it into a format more amenable for the kind of games technology that we use to display it. And there is some kind of trade off of size of the stored data against the speed of display. Roughly speaking we ended up with three levels of detail that you could have. This is part of our test to make sure that it actually showed the right thing at the right time. So we stood there – anybody recognize that? The cobbler – splendid well done. So we stood at the bottom with our laptop, our camera and our GPS and took that photograph and then printed out exactly what we saw on the screen in high level of detail, medium level, and low level of detail. Sure enough I looked at that and said – ‘Yeah, there’s the stream, coming down here’….the sooner you get trees on the maps please that would be very much appreciated…but you can see the shape of the cobbler quite adequately in the data that we had. It’s a little less clear in the lower level of detail, but if we need to that’s actually perfectly good enough for the purposes that we were using. It depends how powerful a laptop you have. I’ll show you demonstration in a minute if I may. I can just about manage medium level of detail on my laptop. Liam had a far more modern thing- students have far more money than the lecturers- so he can do the high level on his, but the lower level would do. Just for reference, we used the Landform dataset profile 1-10k for the dm and the 1-25k for the colour rusted map. If you’ll permit me one slide of computer science in here, what we were doing is checking the amount of disk space we needed to store all this stuff against quality turns out that for the medium resolution, the whole of the UK, we got it to just about 25mb of data and that’s what I’m sitting with here today. We managed to fiddle around and get it going fast enough – you’ll see the results in a second.

Does it work? Yes, we can actually display all of this stuff so we verified it and said yes we’re showing the right thing at the right time. Easy – all we did was a climb a mountain and say ‘yes this is what we should be seeing’. Is it usable? Does it actually solve the problem? And of course that’s when we get very concerned about these computer scientists – yes you can write the programme to do all this wonderful stuff, does it actually really solve the problem? Well, at that stage it got fun. We had to go find some tame pilots and try it out. The Scottish Aero Club at Perth came up with the goods, and we did a bit of airborne software validation. So, there we were sitting in the aircraft approaching Perth to land. Now then, those of you with good eyes will just about be able to see the runway there. You might be able to realize that there is a road down there – there we are following the road on the lead-in to the airport. I have the video of this if anyone wants to see it afterwards please grab me and I’ll show you the various bits. But, basically we took it up in the aircraft – and yes, it shows you exactly what you should see out of the front screen which was good enough for us.

We showed the video to various other pilots and asked for their opinion and I’ve taken some of the Anglo-Saxon out of this quote – this to me kind of vindicated the whole lot: “If you’re in the clag and about to die, use this and you live” – at which point I went ‘Yes, this project is a success, thank you very much’. So it was thought to be useful by a variety of pilots, that was one of the senior instructors who came up with that little quote.

Conclusions – the approach basically works. Liam – one of the best students I’ve ever had as a project student with this came runner up in the National Young Software Engineer of the Year competition. He was also runner up in the Ordnance Survey organised Pushing the Boundaries competition. So it was well thought of and I’m very pleased that he managed to get those particular accolades. Where are we going with this? First thing, we’re probably going to add a thing called highway in the sky which allows us to show planned routes on the thing as well so you actually fly down the corridor and make sure the corridor doesn’t point at the mountain. I would love to release this somehow to pilots on open source and just say ‘There you are, on you go, this will help you. This might actually if the worst day happens, you might just have a way out’. To do that I am well aware that I can’t do that with the Digimap data in there I need to find some kind of open data set, some open DTM which is useful for that and then release it with an open source license. We’re working on putting it on ever smaller platforms. We’ve got somebody working at putting it on iphone – not with a great deal of success yet but we’re getting there.

That’s not me that’s the projector gone…my laptop is still going…anybody? It’s lunch time…well far be it from me to ignore the almighty.

Peter Burnhill: With the intervention, clearly we were heading towards a mountain *laughs* but very many thanks.