From: kelly@dkas.enet.dec.com () Subject: INDUSTRY: VR at Digital Equipment Date: Tue, 9 Feb 1993 22:31:14 GMT Organization: Digital Equipment Corporation Greetings, I run the Artificial Life & Virtual Reality Applications Group at Digital Equipment Corporation. Several folks, having seen our listing in the Meckler Sourcebook and having seen the company's stock recently soar as DEC reorients itself to technologies with a future, have asked me to informally explain the work we're doing. Since most of the requests for info came from usenet readers, I thought I'd beg the indulgence of the moderator and drop a line here. Firstly, let me cover myself legally. 100% of the work we do with external customers - in all of the services fields our larger group works in - is done under non-disclosure. The customers we've traditionally dealt with - for the most part multinationals - see VR, A-Life, and the other technologies we deal in (neural nets, inductive systems, video kiosks, etc.) as conveying a strategic advantage to them in competition with their rivals, so we cannot discuss anything proprietary to customers. This has forced us to put together demos to elucidate techniques and methods we use without revealing proprietary info. It's an odd and difficult way of working, from a marketing standpoint. In a way, we're the opposite of a university. We do not publish, because our customers do not want their critical information publically disbursed. And all knowledge is closely held here, with intellectual property rights always a serious legal consideration. For better or worse, I think this unwillingness to discuss details of on-going projects will be the long term trend in VR. Already, most VR conference speakers, for example, use conferences to promote themselves, their companies, and their products without revealing any details of their expensive-to-develop applications. Nothing wrong with this. Probably inevitable, considering the maturation of the VR industry and the fact that large sums are now involved. But I'm sure VR hobbyists and purists will mourn the passing of the garage VR days as large companies begin to take over some aspects of the field. Seeding: Some folks have realized that while DEC's A-Life & VR Group has been building its foundation in practical commercial VR, DEC has also been seeding the vendor world with trained VR people. Half a dozen VR folks have left DEC in the last few years to work in other large companies, or to found companies of their own. Both Xtensory and ERG Engineering were founded by Digit's, for example. First-rate people involved in both instances. So, DEC may be becoming, in a roundabout way, one of the nurseries for the VR industry. The Work Here: Our work in the A-Life & VR Group is in application development, not VR tool development (another group is doing that) or research. We currently sell consulting services, application development services, and training in VR. And, until lately, we have steered clear of entertainment applications, although obviously that's where the money is now (n.b.: Battletech with 100,000 customers a month @ $7/pop). We have some small interest in the creation of internal apps inside DEC, but they don't generate real revenue, so our focus has always been outward, not inward. Although, we can't reveal details or names, I can mention application areas and fields of concentration. An Aside: The fact that magazine article writers and prognosticators generally wonder in print where the applications will be in VR has always been a source of great puzzlement to us. Our problem has never been finding application areas. It has been isolating the most obvious revenue generators - finding the lowest fruit on a tree bowed down with fruit. But then Alexander Graham Bell was told that the only use of his invention was as a device for allowing the bridge of a ship to communicate with the engine room. So, I guess the intellectual descendents of those foresightful folks are still around and are now wondering about VR applications. We categorize VR applications into three general areas: - synthetic experience (muscle memory skill development) - realization/reification (graphical representation of profuse contextual data) - and perambulation (walkthrough, but not always architectural) In medicine, we're currently dealing with invasive radiology-esque applications. These stretch from: - virtual cadaver creation (Currently in limbo waiting for Alpha and Denali to give us better voxel-level volume rendering for MRI-based cadaver re-creation - We're using MRI rather than waiting for the Skandha and DC-based projects to finish microtoming their bodies.) - to endo/laparoscopic perambulation (Except for finding the textures to represent wet epithelial tissue, endo perambulation is a relatively easy application area - really just accurate modelling and position triggering of features: inflammation, necrotic tissue, lesions, neoplasms...) - to cancer cytogenetics (Especially the modelling of tumour growth and interference in that growth thru site-directed mutagenesis (anti-sense nucleic acid therapy) We're treating homeotic and oncogenic processes as software malfunctions and modelling antisense codon-insertion gene therapies by approaching them as garbage-insertion into cellular utilities or the cells' operation systems. This is the first instance we've ever heard of anyone seriously combining Artificial Life techniques with VR display. In the future, I think all VR will be looked at as just 3-D graphics in a headmount unless it incorporates A-Life for behavioural animation - just an opinion here - but one that is directing our entire approach to VR/A-Life.) - to Medical infomatics (This combined with neural nets, imaging, full motion video): a large complex difficult problem that we've tried to simplify, sometimes using some crude tricks like color-coding (everything yellow: urology and renal, everything brown: procto, everything red: hemotology...) In Telecomm/Utilities, we've had some success in complex network representation. Much of our work in graphical context-heavy profuse data display is really a kind of combination of Visual Languages with Scientific Visualization with VR: bump and terrain data maps, evolutionary sequences, 3-D cellular automota, L-systems for prediction, hydraulic models. This is a vast field and the potential for future development is colossal. At the moment we're exploring faster-than-realtime failure cascade prediction for large energy transmission nets and auto-load compensation for telecomm. On the other side of Telecomm/Utilities have been training simulators and device-use simulators. (Sorry, I know that sounds obscure. Have to use your imagination. Lawyers are involved.) In architecture we've been doing edifice prototyping linked to a spreadsheet so design considerations can be visualized and their financial outcomes viewed simultaneously - and then changed in realtime to satisfy design vs finances constraints. We've also built in environmental impacts of design decisions - allowing a client to view effects of building decisions on purse and environment while walking through the building. In chemical apps, we're looking at the problem from both ends of the spectrum. At one end is manufacturing plant walkthrough/touchthrough (also a project in suspension while DEC's muscle hardware is being distributed - the models are just too large even at the grossest geometry level for PC cpus). At the other end, well along now, is leachate plume analysis (aquifer carriage of toxins, rotting underground storage tanks, Kissimee River redirection) recently successfully demoed to a member of Clinton's transition team. Our most recent interest has been in entertainment applications and in long distance graphically mediated consultation. We're slowly in the process of exploring bandwidth conditions for simultaneous V-world entry thru the net from Germany, Japan, and Massachusetts. Our Findings: One thing we've found in all this work is that pure VR applications seem to be relatively few. Everything we're doing involves either coordination with or direct integration to other technologies. Behavioural animation, for example, involves - genetic algorithms to generate the trial and error sequences that create behaviours of virtual objects, - neural nets (especially boltzmans and Kohonen-style competitive filters, not backprops) to remember consequences of generated actions, - production rules for gene translation, - automata for pattern constriction, etc.). VR, in commercial applications, becomes one technology among many. I think that, in time, as VR entertainment develops, these same concerns will become important to game-builders as well, and they'll have to broaden their scope from just 3-D graphics in a helmet to genuine artificially enlivened universes. Our Strategy: We are integrators - not just of various VR peripherals, or VR hardware with VR software, but integrators of VR with A-Life, nets, realtime process control systems, and everything else already installed and working in commercial shops. We don't expect any customer to toss out their present systems and adopt VR. We make the best use of VR in any particular established MIS/R&D/Production context. We try to use gear from as many vendors as possible, running the whole range from Mattel to Sony. All CPUs until now have been PC-level (workstation level in another group) - which has very obvious limitations. But this will soon change as we convert to Alpha and compete in the high-end graphics market. In SW tools, we've concentrated on Dimension's Superscape, StrayLight's PhotoVR, Sense8's WTK, an internally developed tool, a large company's dev tool, and, in future, Vream. (As an aside for the UK readers: DEC is working with the BBC and Dimension to convert the Cyberzone game to the Alpha platform (world's fastest chip) - most likely this will end up on the Kubota graphics chip too. So, if the Cyberzone TV show lasts for another few months, the update rate should suddenly and drastically soar in a future episode.) More Findings: In the last year, as the group has been building its technical strengths and laying the foundation for what will become a significant business, we've tried to deal with as many vendors and labs as possible in order to determine which we can work with in future and which we can't. The hands-down winner in the impossible-to-work-with category was a company until recently lionized by the press, but now out of business. Their collapse was no surprise to anyone here. There are two others we've found extremely difficult (one sw, one hw) and expect to see them disappear in the next couple years as well. But we've also had some delightful experiences. Hands-down winner in the best-to-deal-with category has been Bruce and Evan at Virtual Research (flight helmet). The most accommodating people have been Beth and the Audio Display Lab rastas at Nasa Ames. And far and away, the best support and most professional assistance has come from Ben Discoe, Ken, Tom, and Pat at Sense8. The best conference so far, I think, has been Meckler 92 in San Jose. And the worst the SigGraph 92 VR day - positively dreadful. Sure hope they overhaul that for next year. Most interesting research, at least for this group, is Carrie Heeter's work up in Michigan, and Randy Pausch's work in Virginia. On a tangential tack: As the number of people we've put through our large variety of applications has reached several hundred, with more every week, and we have a good deal of practical development experience now under us, I've been able to see trends in the behaviour of people inside virtual worlds. We've also had a chance to see some of the dark side of VR thru participant behaviour - something some VR folks eschew discussing (for job security, investment, and other financial reasons). No sense wasting this unique experience, so I've been putting together a book on the future and social consequences of VR based on what we've discovered here. (Yes, I realize. Everyone has a book out or is writing one.) But response to the first several chapters has been very gratifying so it should be out by late summer. Working title is "The Future Of Virtual Reality: Social Consequences Of A New Technology". (My last book is still in print, but won't be much longer, so I need another one out there if I'm still to call myself an author I reckon.) When I have advance order info for the book I'll insert it here (if that's okay with the moderator). On another tangent: One thing we hear a LOT of is concern for the safety and health effects of VR hardware. This is often the FIRST question customers ask. With the recent Nintendo epilepsy warning, the revelations about car phones and cancer, the link between VDTs and fundal cancer (fundus is the back of the eye that the retina maps onto), and the publication of books like "Currents of Death" and "The Body Electric" (discussing deleterious effects of nearby electrical current on cells) it's probably natural that people ask these questions. But I have not heard ANYONE in the VR community seriously address them. I'm sure this is because money is involved - but I know this issue is going to come back to haunt or kill the VR industry if we don't start taking it seriously now. The most I've ever heard mentioned on the subject is a begrudging, "The health effects of this technology have not been studied." I've mentioned my customers' concerns to other folks working in VR, but they've never heard these concerns expressed in their own groups. I think the reason we hear them here is because we deal directly with customers. We're not a school - so we're not dealing with uncomplaining grad students, and we're not a lab - so we're not dealing with uncomplaining techies. We deal with the public, and the public has been hosed so many times by people selling dangerous technologies that it is now rightfully distrustful. We've found that this suspicion translates into real problems that the VR world is going to have to face sooner or later. One principal health concern of customers has been trackers, the other has been headmounts. This obviously has serious implications for those commercializing VR, as we are, because customers have not allowed us to use certain kinds of equipment because of potential health risks. If anyone out there is studying this, I'd like to hear from you on your findings. Why us?: I think that many of the areas we work in, and many of the concerns we have about the technology, have seldom been dealt with by the VR community till now because we're really breaking new ground here in this group. We're not an internal development group. We exist to generate revenue for the company. We're dealing with customers who have serious revenue concerns, who are not interested in technology for technology's sake, and who want to see VR applications in very specific areas that address very specific industry problems and utilize all of the other technologies they've already invested in. These people tend to avoid contacting universities (because they see colleges as places of "research" rather than "application development" - we've seen this countless times). So, such folks seem to come to us instead. Our customers are also a new group of people that VR folks, for the most part, have not yet had to deal with. They are not researchers; they are pragmatic problem-solvers looking for tools and techniques already tested to incorporate into their businesses. And they don't see VR in a vacuum; they want all the technologies necessary to solve their problems all working together - nets, A-Life, imaging, video, long-distance comm, etc.. This may be another reason they come here and not elsewhere. I think these people represent the eventual market for industrial VR (excluding entertainment). So, rather than poohpooh their concerns about health or their wish to mix technologies in applications, we're taking them seriously and working according to their expressed interests. I think we're pretty lucky here to deal with aspects of VR now that everyone will have to deal with later (and not without an enormous amount of resistence from some VR folks, I'm sure). Sum: Hope this little summary satisfies the folks who asked for an update. Virtually all of our work has been in application development, and almost none in discussion or promotion of the work in the last year, for several reasons. Firstly, we have a huge installed base to contact first about VR applications before letting the general public know what we're doing. Secondly, we needed to reach internal groups and sales groups before advertising outside, so the sales folks would not be surprised by customer requests for VR. And, thirdly, we didn't want to be in the position of promoting VR here without actually having done considerable realworld work in the field first. But, as the Alpha machine becomes more available, you should expect to see greater promotion of DEC VR to the outside world, especially if you're already a DEC customer. Anyway, perhaps this was of some remote interest to people keeping tabs on VR's evolution into the commercial world. Richard Kelly Artificial Life & Virtual Reality Applications Group D.E.C. 111 Locke Drive Marlborough, Mass. 01752 kelly@laotzu.dec.com