Back in '95, working for a DoD contractor, I saw a demo of some augmented reality gear being proposed for equipment troubleshooting and maintenance applications. The goggles were the same material used in head-up displays, coupled with an earpiece/microphone/camera in the headgear.
The microphone provided input for possibly the best voice input setup I've ever seen to date, in that it actually recognized a user saying things like, "yeah, uh-huh" and similar grunt/groan acknowledgements. The goggles were linked to a rather cool pattern recognition system. A small status bar in the side of the display provided a couple of icons to let the user know they were on the right track, a smiley/mr. yuck set of icons provided a status as to system "lock" status (or not).
The major downside to all this was the hardware in '95 to support all this was rather intensive. I recall 3-4 towers, plus a custom-built rather hefty cube-shaped box (no, not a next cube) that handled all the pattern-recognition processing and voice input. Oh, and the hefty wiring to the headgear.
Re: Multiple providers, there's something to consider here. Having multiple upstream providers from the hosting service is fine for reducing number of hops to the end viewer. If true redundancy is an issue, you also need to verify that the hosting service has their data lines coming in at multiple entrances to their facility. Having two DS-3 circuits is great, but if the LEC provisions them both on a single OC-3 Fiber entrance to the building, it still only takes one backhoe to kill them both.
Another question you need to ask, who are the upstream providers? If your hosting company is buying upstream from tier one backbones (UUNET, Sprint, etc.) that's one thing, but if their upstream bandwidth is coming from Joe's Bait-n-Tackle, ISP, and lumber yard, you can bet there are going to be problems down the road.
In the rush to verify server reliability and redundancy, don't forget to look at the network engineering aspect...
Slashdot Mirror
Back in '95, working for a DoD contractor, I saw a demo of some augmented reality gear being proposed for equipment troubleshooting and maintenance applications. The goggles were the same material used in head-up displays, coupled with an earpiece/microphone/camera in the headgear.
The microphone provided input for possibly the best voice input setup I've ever seen to date, in that it actually recognized a user saying things like, "yeah, uh-huh" and similar grunt/groan acknowledgements. The goggles were linked to a rather cool pattern recognition system. A small status bar in the side of the display provided a couple of icons to let the user know they were on the right track, a smiley/mr. yuck set of icons provided a status as to system "lock" status (or not).
The major downside to all this was the hardware in '95 to support all this was rather intensive. I recall 3-4 towers, plus a custom-built rather hefty cube-shaped box (no, not a next cube) that handled all the pattern-recognition processing and voice input. Oh, and the hefty wiring to the headgear.
Other than that, pretty darn cool back in '95...
Re: Multiple providers, there's something to consider here. Having multiple upstream providers from the hosting service is fine for reducing number of hops to the end viewer. If true redundancy is an issue, you also need to verify that the hosting service has their data lines coming in at multiple entrances to their facility. Having two DS-3 circuits is great, but if the LEC provisions them both on a single OC-3 Fiber entrance to the building, it still only takes one backhoe to kill them both.
Another question you need to ask, who are the upstream providers? If your hosting company is buying upstream from tier one backbones (UUNET, Sprint, etc.) that's one thing, but if their upstream bandwidth is coming from Joe's Bait-n-Tackle, ISP, and lumber yard, you can bet there are going to be problems down the road.
In the rush to verify server reliability and redundancy, don't forget to look at the network engineering aspect...