Re: Geolocation and POI

Hi all,

When we organized the first ARDevCamp last year, it was because we saw that
AR, geoweb and 3D web folks were about to collide, and we felt it was
critical to get them talking about open standards. It's a very difficult
discussion, and our ideal was for AR to emulate the openness and
extensibility of the web, so I'm delighted to see W3C in the mix. (For
clarity, "we" were myself, Mike Liebhold, Damon Hernandez and Chris
Arkenberg).

That said, I'd like to explore the assumptions behind this working group
proposal, if that's ok. From what I can see, there was discussion at the
W3CAR meeting about the desirability of a standard representation for "POIs"
across AR platforms, which led to the proposed POI WG. How broad or narrow
is the group's conception of POI here? For example:

1. Are we talking about the very specific notion of a single point in
3-space, fixed within some geodetic coordinate reference system (CRS)? This
is I think the generally accepted notion of a POI, although in many cases
POIs are defined only in 2 dimensions today.

2. In this discussion, do we include more complex geometries beyond points?
I note that KML and the proposed variants ARML and KARML are mentioned; KML
already goes well beyond single points, to include 2D lines & polygons for
instance. For AR, I suspect we would also like to represent 3D spatial
geometries (not object models, but 3D regions of space), would this be a
valid concern of the WG?

3. Do we include non-geodetic coordinate systems in the discussion? X3D for
example supports geocentric systems. We can also imagine AR use cases for
heliocentric or larger CRSes (an AR model on the solar system).

4. Broadening further, what about non-numeric spatial identifiers such as US
zip codes, Yahoo WOEIDs, place names ("Heathrow Airport", "San Francisco"
etc)? There's a lot of interest and expertise in this area in the Geo
community; is this envisioned to be part of the "POI" discussion?

5. In AR, things move over time. Does the proposed WG contemplate temporal
paths? Or only POIs that do not move relative to an earth-based coordinate
system?

6. In AR, relative positioning is extremely important. I realize this moves
well beyond the common definition of POI, but there could be significant
value in standardizing how these work in a web context. Is this in or out of
the discussion?
- Specifying a location relative to the client device is one use case (e.g.,
a graphical windowing system in my personal AR space).
- Specifying location relative to a defined reference point (e.g. fiducial
markers) is another.
- Specifying location relative to an object is a third case ("Alice’s avatar
is on a boat”. “Bob’s tweets float above his head as he walks by”. “Charlie
is wearing a virtual hat”. “The virtual spider is on the virtual table,
which is on the physical floor”).

So, how broad or narrow do we think the scope of this "POI" WG is? If it
only seeks to agree on the simplest case of points and polygons that are
fixed in earth-based CRSes, then I'm all for putting it into the Geo WG.
Getting an agreement on these basic things will be useful to AR, the Geo
guys have done a tremendous amount of work already, and reinventing the
wheel seems pointless. It won't address most of the AR-specific issues, but
it would be a start.

If the goal is to produce location standards that are more broadly useful
and responsive to the particular needs of AR, then it should probably be a
separate WG with an AR-specific charter beyond "POI". I guess that
discussion should also look to existing W3C 3D efforts; Web3D folks are
becoming active in AR, so there could be some converged interests here as
well.

Of course, as Christine points out, there are many more "triggers" for AR,
beyond just location. I'm not sure if these are ripe for standardization,
thoughts on this?

There is also the issue of data representation in AR; I'd like to think that
a mechanism involving something analogous to user-agents and MIME types
could help us get to client-aware adaptation and data extensibility.

Then I suppose there's the question of what the AR equivalent of a DOM would
be...but now I'm rambling. Perhaps an IG is a good idea, regardless what we
think about the POI WG proposal?

Best,
Gene Becker
@genebecker

On Fri, Jul 9, 2010 at 9:34 AM, Dan Brickley <danbri@danbri.org> wrote:

> On Fri, Jul 9, 2010 at 6:11 PM, Christine Perey <cperey@perey.com> wrote:
> > Hello Lars Erik,
> >
> > Thank you for your invitation to inputs on this topic.
> >
> > As you may realize, AR POI involves location and device orientation as
> *two*
> > of the many possible types of metadata attributed to a point of interest.
> >
> > The "point" can and will very frequently, be an object (thing or a
> person)
> >  moving in space/without geoposition or orientation associated. Even if
> the
> > GEO WG charter were to include feature extraction attributes, I fear that
> > the AR community could have difficulty "seeing" the definition of POI as
> > their domain for contribution if it is "housed" in the GEO WG.
>
> To offer a balancing viewpoint, in my experience the AR community are
> very well aware that their 'technology' is a hybrid composed from a
> wide range of standards and potential standards. It's also common to
> find 'the map is not the territory' observations amongst AR
> technologists. If a restaurant or a house for sale pops up in 3D AR
> view, AR people are plenty smart enough to know that it is *not* their
> core business to enumerate properties like 'housePrice' or 'vegan?',
> even if that kind of information will ultimately prove critical to
> their applications. AR, as the name suggests is a bridging technology;
> many kinds of description and identification will be used, from
> barcodes, NFC, face and audio recognition, ... Some of these will be
> standardised elsewhere, perhaps some at W3C, and others will be
> value-adding facilities associated with particular apps. The whole
> system will evolve over time. Frankly, if "the AR community" are
> unable to find a way to contribute to POI efforts just because they're
> in a GEO WG rather than an AR WG, that should be a huge alarm bell for
> anyone at W3C considering chartering a group for them. I'm much more
> optimistic though...
>
> > My major concerns are that
> >
> > (1) the AR work will be lost (not receive the attention it warrants) if
> it
> > is part of the Geo location group.
>
> I think you underestimate the intelligence of the AR community here.
>
> > If the GEO WG scope/charter encompasses all the possible AR activity,
> these
> > will gradually become "charter" creep.
>
> "All possible AR activity" includes describing opening hours of
> restaurants, face recognition, barcodes in all their flavours, audio
> fingerprinting, category systems, movie lookups, voice (hands free)
> browsing, ... ... pretty much anything. AR is open-ended by definition
> (the "reality" bit). The idea that W3C might try to stuff all such
> activities into a GEO WG is ... unconvincing.  Nobody at all is
> proposing this!
>
> > (2) people who originally chartered the GEO WG are not the best to be
> > implementing/writing the recommendations which need to be implemented by
> the
> > AR platform publishers.
>
> Which people are you talking about? why do you expect the initial
> drafters of the Geo charter to be the only contributors to future
> work?
>
> > My proposal is that there be two separate WGs within W3C and that these
> have
> > close working relationships and perhaps co-locate a meeting per year, as
> > needed/ convenient.
>
> I'd suggest a single GEO WG take on the task of describing basic POIs,
> and that its charter require Web-style extensibility that will allow
> multiple other parties to extend those descriptions in unbounded ways.
> I suggest W3C also charters some AR Interest Group whose role would be
> to broker standards-related  collaborations amongst the many different
> technical aspects that make up the AR application landscape.
>
> The money and time saved from not having 2 groups working on the same
> thing can be re-allocated to other liaison activities (assuming that
> the money/time exist in the first place).
>
> > A separate WG will permit the charter to treat AR is a "mash up" of many
> > technologies. An AR WG will have the ability to establish liaisons with
> many
> > groups which includes Geo WG but others which are outside the scope of
> > interest of a GEO WG.
>
> AR is indeed a combination of many technologies! An AR IG could as you
> suggest establish many liaisons with related groups.
>
> cheers,
>
> Dan
>
>