Darren.Hague@brunel.ac.uk (Darren S Hague) (03/08/91)
Just a response to the call for people to say what research they're doing. I've just started my PhD at Brunel University, where I am looking at architectures for VR. I am currently working on a layered architecture: +----------------------------------------+ | I/O Servers (Users' headsets & gloves) | +----------------------------------------+ | Actor Processor | +-------------------+--------------------+ | Object Forms | Object Semantics | +-------------------+--------------------+ | World Processor | +----------------------------------------+ | Environmental Laws | +----------------------------------------+ The Actor Processor mediates between the user interface (for multiple users) and the representation and viewpoint of the Actors as objects. It also allows access to the Object Forms (ie graphics info) to render the environment from the viewpoint of each actor. The World Processor mediates between the Laws of the Environment (such as gravity, Newton's Laws, etc), and the objects they apply to at the Object layer. This must be done in real time. Each object is represented by two interrelated notions of Form and Semantics. An object's form is simply its graphical information (eg position, colour and radius for a sphere), whereas the semantics holds abstract information such as mass, velocity, relations to other objects and so forth. These notions are seperated in the architecture such that the scene renderer needs only to look at the graphics database, and can safely ignore the information in the semantics database. The architecture is very close to that of a 2nd generation knowledge-based system operating in real time, and I will be searching literature in Knowledge Representation, Real-time Simulation, Real-time Techniques, Space-time coherency in Graphics Representations and General Systems Theory. At the moment, I have been able to find *no literature at all* on VR from an info. systems point of view, only HCI stuff concentrating on hardware. Any references gratefully received! I hope this is of use to others in the VR research community, and I look forward to hearing from you all. Cheers for now, Darren Hague Darren.Hague@brunel.ac.uk
jay@uunet.UU.NET (Jay Nelson) (03/09/91)
In article <18015@milton.u.washington.edu> Darren.Hague@brunel.ac.uk (Darren S Hague) writes: > > >Just a response to the call for people to say what research they're doing. > >I've just started my PhD at Brunel University, where I am looking at >architectures for VR. I am currently working on a layered architecture: > >+----------------------------------------+ >| I/O Servers (Users' headsets & gloves) | >+----------------------------------------+ >| Actor Processor | >+-------------------+--------------------+ >| Object Forms | Object Semantics | >+-------------------+--------------------+ >| World Processor | >+----------------------------------------+ >| Environmental Laws | >+----------------------------------------+ > >The Actor Processor mediates between the user interface (for multiple users) >and the representation and viewpoint of the Actors as objects. It also >allows access to the Object Forms (ie graphics info) to render the >environment from the viewpoint of each actor. > >The World Processor mediates between the Laws of the Environment (such as >gravity, Newton's Laws, etc), and the objects they apply to at the >Object layer. This must be done in real time. I'm playing around with the ideas of virtual worlds at home on a PC. I wnat to see if I can build a method for "flying" through raw data files. To understand the concepts, however, I started with the simpler problem of having a fixed room of objects and "flying" around the room. The architecture I'm trying right now is a little different: +-----------------------+ +---------------------------+ | User inputs | | Display outputs | +-----------------------+ +---------------------------+ | Environment I/F | | View translation | +-----------------------+ +---------------------------+ | Virtual pilot I/F | | Virtual View | +-----------------------+ +---------------------------+ | ^ | | V | +-------------------------------------+ | Virtual Space | | - Pilots | | - Objects | +-------------------------------------+ The virtual pilot I/F describes all the positioning, motion, and point of view information for a pilot, irrespective of the physics being modeled. The environment I/F constrains access to pilot commands to fit the physics of the modeled environment. The layering allows the user interface to be replaced (I have only a keyboard now, but will probably get a mouse later) but requires any new interface to be constrained by the same physics. Changes to the virtual pilot are made using an internal clock. The internal virtual world consists of pilots and objects. The virtual view is the part of the virtual space visible to the virtual pilot. This space is translated to the display view before it is drawn. Notice that more than one virtual pilot could exist and they could see each other if necessary. I need to have a pilot view from the ceiling that can track the pilot the user is guiding throught the room, so that the user doesn't get lost. My architecture is driven by flexibility in user input devices and display translation (when flying through raw data I want the user to define what it will look like). I'm also currently assuming that objects don't move since I will be viewing static raw data, although they could be pilots rather than static objects. >Each object is represented by two interrelated notions of Form and Semantics. >An object's form is simply its graphical information (eg position, colour and >radius for a sphere), whereas the semantics holds abstract information >such as mass, velocity, relations to other objects and so forth. These notions >are seperated in the architecture such that the scene renderer needs only >to look at the graphics database, and can safely ignore the information in the >semantics database. I like this idea. It maintains independence of view and behavior information. In my architecture the Semantics would get interpreted in the virtual space, but the Form would be interpreted in the view translation. >I hope this is of use to others in the VR research community, and I look >forward to hearing from you all. > >Cheers for now, >Darren Hague >Darren.Hague@brunel.ac.uk Thanx. It is useful to compare other approaches. I would welcome any comments on the approach I am using. Most of the articles I've read here have been more wishful thinking than concrete approaches. Jay Nelson (TRW) jay@wilbur.coyote.trw.com From news@minyos.xx.rmit.OZ.AU Fri Mar 8 19:12:20 1991 Received: from munnari.OZ.AU by milton.u.washington.edu (5.61/UW-NDC Revision: 2.1 ) id AA13742; Fri, 8 Mar 91 19:12:15 -0800 Received: from werj.xx.rmit.OZ.AU by munnari.oz.au with SMTP (5.64+1.3.1+0.50) id AA25573; Sat, 9 Mar 1991 13:12:08 +1000 (from news@minyos.xx.rmit.OZ. AU) Received: from minyos.xx.rmit.OZ.AU by werj.xx.rmit.OZ.AU with SMTP. (5.65a) id AA05119; Sat, 9 Mar 91 14:11:44 +1100 (from news@minyos.xx.rmit.OZ.AU for sci-virtual-worlds@munnari.OZ.AU) Received: by minyos.xx.rmit.oz.au Newsgroups: sci.virtual-worlds Path: godzilla!mg From: mg@munnari.oz.au (Mike Gigante) Subject: Artists/VR Organization: RMIT Computer Centre, Melbourne Australia. Date: 8 Mar 91 03:03:59 GMT Message-Id: <mg.668401439@godzilla> References: <17998@milton.u.washington.edu> Sender: @minyos.xx.rmit.oz.au Apparently-To: sci-virtual-worlds@munnari.OZ.AU Status: RO I may send more details later, but note about the German conference motivated me to mention our interest in VR & Art. (tho' the motivation for the work is not just Art, it is using artists to push back the bounds of the technology) 1) VR Sculpture, this project includes an Australian sculptor Robert Owen (he is also head of sculpture here at RMIT). Robert and I have also put a VR proposal in for the siggraph '92 "Portal" 2) VR Performance Art, involving the artist STELARC. Stelarc is best known for his high tech performances that combine electroic instrumentation, a third hand and video We also are interested in VR Architectural design. Mike