Open AR Panel – AWE 2013
June 5, 2013, Augmented World Expo, Santa Clara, CA—The AR infrastructure needs to move to standards and open platforms to become as recognized and used as USB. Christine Perey from Perey Consultants moderated a panel of people working on those standards. Panelists were Rob Manson from ARStandards.org, Neil Trevett from Khronos, and Martin Lechner from Wikitude.
Perey opened with the need to change to open and interoperable AR. These changes have to work in all phases of the industry, from physical, to authoring, to delivery to result in positive user experiences. Like the Web, AR should work on any platform, anywhere, from any software, workflow, and client. The issues against these changes are compute complexity, fragmentation, and the existing technology and content silos.
As a result, the whole industry needs a driver company or organization. The on-going works include existing standards from organizations like MIPI (Mobile Industry Processor Interface), OCG, Khronos, IEEE, ISO (International Organization for Standardization), MPEG (Motion Picture Experts Group), NISO (National Information Standards Organization), W3C (World Wide Web Consortium), Web3D, ITU (International Telecommunications Union), OMA (Open Mobile Alliance) and SID (Society for Information Display). Much of the work is being driven by the ARStandards.org group. This group is working to coordinate efforts, create resources, and help in identifying overlaps.
Trevett described the ongoing work and existing standards from Khronos. The rationale for standards is that it makes the base functions easier to integrate. Standards have to be timely and need relationships for a relative consensus. Final details are hashed out in committees.
For APIs they have released Open GL, GL-ES, CL, and VX. The latest versions have handles for stream inputs and camera control. The programming APIs are in C, but are being ported to Java. The functions will also be in Collada. There is some work on moving the graphics functions into HTML5 and to native code for some processors.
Manson noted that the standards are moving into the AR field. Currently, AR is just another feature. WebRTC is one of the organizations that is working on identifying and creating real-time communications functions for browsers through Javascript APIs. They have defined layers for overlays, video, etc. and the associated protocols for accessing those pieces.
Most of the latest browsers are adding the necessary internal functions, but Apple is lagging. The web-based framework in Javascript is working towards consistent data models. The ongoing evolution of AR technologies is moving towards a convergence on Web platforms that use WebRTC for peer-to-peer communications that can share cameras and sensors. Still, the processing pipeline is inefficient.
Enabled browsers can track and detect features within a frame, and will become integrated with WebRTC and visual search for many functions. The browsers will also integrate Web GL/ GL-SL for GPU acceleration, and will access other Khronos standards as applicable. More information is also available from the W3C AR group.
Lechner provided an overview of the language for AR functions. The standards are addressing the scope of AR in a mark-up language (ARML) on mobile devices. The visual functions consider the camera and display characteristics. The language is a descriptive format with a script for interactive functions that depends upon existing standards.
The basic structure starts with an object to be augmented. Features on the object become anchors to locate the object in the real world, provide geospatial information, generate markers, and enable feature extraction. These functions are linked to other assets that are delivered to some visual representation. For more details, see www.opengeospatial.org to see the details of the ARML 2.0.
Define open AR?
Lechner responded that the issue is too broad. For interoperative AR, you have to define the use case and platform.
Trevett agreed, and added that the emerging adoption of standards is adding value and easing implementations.
Manson extended those responses by noting that there are classes of AR, and large differences between mobile and fixed applications. There are many types of client devices and software.
Browser, clients, AR apps?
Lechner claimed that there are differences between AR browsers, so a common interface enables easier upload to general browsers. In comparison, apps have a restricted feature set per problem of function.
HTML in the Web, what is the tradeoff in performance versus a native app?
Manson noted that an AR browser still has universal access, but currently there is no interoperability between AR and standard browsers. The issue is one of transformations and transcoding to integrate the AR functions.
Lechner commented that the ARML initiative is working to support a similar feature set with a common format on any AR browser. These changes are to overcome the limitations of existing proprietary AR engines. Long tern, we will see everything move to the Web. Adoption of standards will eliminate the need to down-load an AR app to use AR.
Manson complained that the biggest hurdle is that consumers have to know of the content to move to get that content. Then they have to identify the engine to view the content.
Trevett noted that AR browsers are moving into standard browsers and many of the latest versions have the internal tools and links to do the AR functions. Other drivers like video conferencing, etc. are enabling faster convergence of capabilities and in-browser is getting close to native app performance.
Content on the Web is easy to redirect. AR needs to identify markers, data sets, etc for function?
Manson suggested that this can be solved by adding a script tool to a website. People will not look directly for AR, but just want it to be there for the information and experience.
Seamless search and content delivery for directions and catalogs?
Trevett posited that Google will eventually have AR search terms
Push from AR industry or Web developers to drive standards?
Manson suggested looking at the demos on ARbuild. Developers will build the content and will drive the standards. Browsers will ship with AR backed in. as long as the hardware and tools like SDKs and APIs are available, the developers will make the content.
Trevett offered open interfaces from platform providers will prevail. For example, AOL tried to control the whole interface, and lost to the general purpose browsers. The industry will move to open interfaces from all platform providers.
Will creators continue to have separate links or URLs to operate AR?
Lechner said that the Android playstore checks for interoperability. Then it goes directly to the playstore. This is a partial solution.
Manson added that this is a transitional issue. The installed base of users are mostly keeping on the leading edge due to automatic updates to their browsers.


