Monday, May 14, 2012

Presentations from GeoAlberta 2012

I spent last week up in Calgary for the GeoAlberta conference, which was celebrating its tenth anniversary. It was a good event and I attended a number of good presentations. Dale Lutz of Safe Software gave an interesting review of the history of geospatial data, and my vintage 1990 paper on Exploiting Relational Database Technology in GIS got a mention in the section about moving into databases.

I gave the opening keynote talk on "Geospatial Everywhere", on trends in the industry. For those who've seen my presentations previously there will be some things you've seen before, but there's also a lot of new material in this version, including quite a bit on the multimedia trends that I wrote about in my previous post. You can see the talk on vimeo here:



I also participated in a panel discussion with Ed Parsons of Google and Chris Moore, CIO of the City of Edmonton. Chris gave an excellent keynote talk earlier that day - I was very impressed at what City of Edmonton is doing in its IT operation, it is definitely one of the most progressive local government organizations that I've come across. We covered a good range of topics on trends in the geospatial industry.

Wednesday, May 2, 2012

Geospatial technology in 5 to 10 years


I have been absent from blogging for a while, working on various interesting new things behind the scenes! I am planning to make an effort to be posting here more regularly - yes I know everyone says that, we'll see if I can manage it!

To kick things off again, the post below is a document I was asked to write for the United Nations Programme on Global Geospatial Information Management (GGIM), which is an inter-governmental mechanism to consult on issues related to global geospatial information. Contributors were asked to write on how they saw future trends in their aspects of "geospatial information management"over the next 5 years, and also looking further out to 10 years. The summary document compiled from all the contributions is here.

It's quite interesting to try to think ten years out, and there's quite a bit of detail I had to cut out to get to 1000 words. I'll expand on some of these topics in future posts and in some of my upcoming talks (I'm doing keynotes at GeoAlberta in Calgary next week, and at GITA Australia in Melbourne in August).

Current trends

The following are important current trends that will have a significant impact on the state of the geospatial industry in 5-10 years:
  • Growth in use of “neogeography” and open source geospatial solutions
  • Increasing use of “multi-media” geospatial data - such as Google Street View, Microsoft Photosynth, and georeferenced photos and videos
  • Increasingly pervasive location tracking
  • Use of smart phones
  • Crowdsourcing for data creation and maintenance
  • Augmented reality
  • Heads up displays

The hardware landscape in 5-10 years

Moore’s law says that computing price-performance will increase by a factor of 10 in five years and 100 in 10 years.  Increases in network bandwidth follow a similar but slightly slower trajectory. In ten years all mobile devices will have GPS, accelerometers and a compass, as most do today, and potentially other sensors that may help with indoor positioning, a problem that is not yet solved at the mass market level (see more below). Devices will also include multiple very high definition cameras.

Video data

In ten years time it is likely that all smart phones (or whatever replaces them) will be able to film 360 degree 3D video at incredibly high resolution by today’s standards, and wirelessly stream it in real time. In addition to this capability being available on mobile devices used by people, cameras like this will exist in a very small form factor at very low cost, so they could be deployed in very large numbers for various applications. It is likely they could easily be powered by small built in solar panels. There would likely be a mixture of public and private video streams from these devices. Individuals might choose to share their streams publicly or with friends, at least some of the time – this is a logical extension of current social network behavior (publishing photos online). Such devices would likely be carried or worn by workers in situations where it would be useful for their colleagues (back at the office or in the field) to be able to see what they are seeing – for example police officers, firefighters, utility workers, etc. They would also be mounted in many vehicles, at street intersections, etc. This network of devices will provide data that can be merged in real time to give an immersive video view of the world, like a Google Street View or Microsoft Photosynth. The nature of this imagery will enable accurate 3d models to be built, at the very least at the point cloud level, but most likely at a much more structured “intelligent object” level. Of course all these feeds would be recorded, allowing viewing of both historical and real time data. This data can be easily correlated with real world objects that have location tracking devices (or with static objects with a known location), and objects can also easily be identified through markings like QR codes.

Augmented reality applications will be pervasive, with the ability to view a whole range of data overlays on top of the real world. In addition to using hand held devices like current smart phones, other likely form factors for AR applications include glasses or goggles, contact lenses and even direct projection into the eye.

Pervasive location and other sensors will provide extremely granular information from vehicle and foot traffic to restaurant occupancy, performance of utility networks and so on. All of this information will be available in real time as well as generating masses of historical data.

Software

The multimedia geospatial data described above will require quite different types of system from today’s geospatial applications and it is likely that new entrants into the market will become major players. Already in today’s geospatial market we are starting to see significant competition for the long established traditional GIS vendors from two, sometimes overlapping, directions: the first is from so-called “neogeography” systems such as Google and Bing Maps, and a host of startups, and the second is from open source geospatial software. Even in a five year timeframe we will see significantly more diversity in the geospatial market than we have had over the past couple of decades. We are likely to see much more influence from video games, in terms of dynamic graphics and 3D visualization – something that Google Earth (initially Keyhole) pioneered in the geospatial space. This will be another driver for a new generation of software to replace today’s incumbents.

Crowdsourcing

Crowdsourcing has already been well proven as a valuable tool for creating and maintaining geospatial data, both active crowdsourcing, of which the best known example is OpenStreetMap, and passive crowdsourcing – using data streamed from location sensors in phones (or elsewhere), both to generate traffic information and to identify changes to a road network or other data. Georeferenced photos and video help substantially with crowdsourcing, especially for data validation. It is hard to see how there will still be a market for datasets like those currently sold by NAVTEQ and Tele Atlas in 5 years time – they will have been superseded by crowdsourced datasets from OpenStreetMap or something similar to it (and using a combination of active and passive crowdsourcing).  Availability of these types of datasets also raises difficult questions for national mapping agencies, who are likely to find their products used in increasingly niche areas, and find it difficult to justify the costs of traditional data maintenance mechanisms.

Indoor positioning

One problem that is surprisingly difficult to solve is that of accurate location tracking indoors. This is primarily because of multipath – sensor signals are easily reflected off walls, floors or ceilings, and so it is easy to get false readings. And also because it is hard for many types of external signal to reliably penetrate buildings. Different technology approaches have different strengths and weaknesses, and it is likely that multiple technologies will be needed to solve this problem with widespread coverage. Technologies that could play a role include ultrawideband, video, accelerometers and RFID. A reasonably pervasive solution is unlikely in five years, more likely in ten years.

Privacy and policy questions

Several of the trends mentioned here raise some difficult issues in regard to privacy – including both specific recording of location of individuals, and widespread recording of video. In general technology is running ahead of policy. Space prohibits a more detailed discussion of this topic here, but it is an important area for focus.