We Begin To Develop Nervegear!

[BrainJackers]

I said “both an understatement and an overstatement” because the way a sword is swung in most video games is actually easier than the robot arm controlling tasks Meng’s study participants were attempting. In a video game, swinging a sword usually requires just one button push to trigger an animated sequence in the game avatar, fewer and less precisely controlled actions than the “4 buttons” in the studies.

Using SAO as a reference point, you need to maneuver your body into a certain place to press a button, but I meant without Sword Skills. To calculate the swing of the arm, any slight movements you have off of it (we aren't perfect), when you stop, the way your fingers are holding the sword, the possibility the sword might slip out... I may be overthinking it but for true immersion with no sequence breaks...

 

--@Kayaba

Edited February 12, 2017 by BrainJackers

[Josua]

I've been monitoring for progress on the possibility of an actual NerveGear in real life. While we do have the tech (while still at its infancy), nobody is trying to make something like that.

 

For it to be a reality, we need to focus on some things:

• Total Immersion

I'm sure everyone by now should have an idea of what level of immersion does the NerveGear, as depicted in SAO, offers. All five senses: Taste, Sight, Sound, Smell, Touch; are all simulated properly by the NerveGear. It's said that it was due to the microwave transceivers that redirected all brain signals from the brain to the game with the device. However, as we all know, microwaves don't do something like that in our reality. As such, we can only rely on two options: EEG & the such, or implants. EEGs, while already commercially available, is not powerful enough to track every single brain signals that are sent through our synapses, only capable of relaying data on the general activity of the brain & several particular sections if need be. Improvements on EEG are necessary, but to the extent of tracking every brain signal that shoots through our synapses every instant, it will require heavy/major improvements on not only detection capability, but also it must be able to manipulate / use the brain signals for the desired goal. And, let's not forget power efficiency, since it is not wise to have a powerful brain-detecting device that takes quite the toll on electricity.

 

For implants, however, this is a different issue. Safety for prolonged usage, compatibility with user's tissue/body, ease of removal & re-usage. These are quite the issue to tackle with modern tech. While we have made great strides in the tech that supports implants, the implants themselves are still an issue. The compatibility with the user's tissue/body & ease of usage & removal, these are still a problem, since most implants, at the moment, can damage the users, regardless of compatibility. Improvements must be made on the implants so that it won't reject the users & cause any damage that may be lethal to the users.

 

• Rendering & Processing Power

As per the described specifications in the SAO Wikia on the NerveGear, it uses a diamond semiconductor central processor. While using diamonds for processors are already a proven concept, it is still expensive, since making it would be a tad expensive. However, as a substitute, we can look at a graphene-based processors. While it is still too expensive to manufacture graphene, however, it is already a proven fact that it is way faster than any CPU available in the current market, save for quantum computing ones. Assuming that it could be developed within 10 years to be easily manufactured with a feasible price, it could prove to be a game-changer.

 

As for cooling, we do know that despite not having any cooling fans or any form of cooling devices, it could still function without any drop in performance. As such, we need to consider an internal cooling method that will be able to be maintained for a long time, since it's not logical to use watercooling or cooling fans.

 

Rendering would be a monster to deal with, aside from the CPU. Rendering the level of realism offered by the NerveGear is practically near-impossible with current generation GPUs. Because of this, the GPU needs not only to be compact & power-saving, it must also be powerful enough to render that much pixels.

 

• Connectivity

Well... This is not really much of an issue, although it is still an important aspect of the NerveGear. Aside from LAN Connection, WAN is also used in lieu of any LAN cable. However, if we're talking about what if it's suddenly disconnected, it would be a huge problem, if we're on the original NerveGear. However, with our own iteration, we won't have to worry much about it, since we won't die if we suddenly get disconnected from the internet connection.

 

• Power Source

Now this one... Is a tricky one. The NerveGear is powered by two means, Battery & Direct Charge. Direct charge can be through the charger of the battery, while charging the battery, it is also powering the NerveGear without using the battery charge. If it were to be unplugged, the NerveGear would function using the battery as the power source. The issue would be the size of the battery & the amount of power it can hold.

 

As per described in the anime/manga/LN, most of the NerveGear's weight comes from the battery. It can be said that it is quite the size. However, the amount of power it can hold is while debatable, can still be considered to be reliable, since the NerveGears of the players when they were transferred from their homes to hospitals were still functioning. However, this kind of reliability is somewhat scarce to be found in our reality. Batteries that are usually Lithium-Ion aren't really that reliable if we're talking about number of charges before degradation of performance. The alternates are many for the battery, though I'm only aware of some. Lithium-Potassium is considered to be the replacement for the aging Li-Ion battery, with better power storage & amount of charges. However, not many have caught up with the Li-Po battery, resulting in the still wide usage of Li-Ion batteries. Also, graphene-based batteries are probably the next breakthrough. Capable of more charge numbers than Li-Ion & Li-Po batteries, it is also capable of storing more power than both. However, graphene is still expensive to manufacture in mass production scale. Until a new method of manufacturing graphene at a low cost is available, graphene batteries would be considered a secondary option.

 

This is all that I can say on the NerveGear creation issues/solutions at the moment. Although I probably didn't include some issues/solution, I hope that this might give you guys more things to talk about.

[BrainJackers]

However, as we all know, microwaves don't do something like that in our reality.

 

They are a form of electricity that could be used to stimulate the brain. It would be horrible and something such as Infrared is much better, but they can do what was described.

 

As such, we can only rely on two options: EEG & the such, or implants. EEGs, while already commercially available, is not powerful enough to track every single brain signals that are sent through our synapses, only capable of relaying data on the general activity of the brain & several particular sections if need be. Improvements on EEG are necessary, but to the extent of tracking every brain signal that shoots through our synapses every instant, it will require heavy/major improvements on not only detection capability, but also it must be able to manipulate / use the brain signals for the desired goal.

 

EEG is fine; EEG products aren't. Beyond that, EEG isn't to manipulate brain signals. That's what something like Infrared energy/Magnetic fields would do. Implants are able to do both, just as CPU can take input and deal output throughout its many pins. 

 

For implants, however, this is a different issue. Safety for prolonged usage, compatibility with user's tissue/body, ease of removal & re-usage. These are quite the issue to tackle with modern tech. While we have made great strides in the tech that supports implants, the implants themselves are still an issue. The compatibility with the user's tissue/body & ease of usage & removal, these are still a problem, since most implants, at the moment, can damage the users, regardless of compatibility. Improvements must be made on the implants so that it won't reject the users & cause any damage that may be lethal to the users.

 

Implants are risky but there's no real compatibility issues with their tissue. Possibly, the person might have chemicals that would harm an implant or not have the tissue where the implant would be placed though.

 

Why would you remove or reuse it?

 

When has an implant rejected the user?

 

Most implants are done for medical reasons and focus on new technology. Any implants used would be standardized and thoroughly tested before hand. I feel saying it must be improved for safety ignores the fact they're prototypes. 

 

The main risk is the surgery itself. That surgery is also a turn off for any potential consumers.

 

However, as a substitute, we can look at a graphene-based processors. While it is still too expensive to manufacture graphene, however, it is already a proven fact that it is way faster than any CPU available in the current market, save for quantum computing ones. Assuming that it could be developed within 10 years to be easily manufactured with a feasible price, it could prove to be a game-changer.

 

Graphene was in the news a few years ago and has had no updates since. Graphene CPUs also very fragile.

 

Rendering would be a monster to deal with, aside from the CPU. Rendering the level of realism offered by the NerveGear is practically near-impossible with current generation GPUs. Because of this, the GPU needs not only to be compact & power-saving, it must also be powerful enough to render that much pixels.

 

 

Roughly 180 1080s, perfectly synced, are needed to render Doom Ultra Level Graphics. I'll post the math (which was rough and not at all to be used in a production environment) for that in a bit. I'm away from the paper itself. 

 

You still have yet to cover figuring the brain out (mapping senses and figuring out an individual's neurons), storage, servers, hijacking the body.... 

 

An atomic hard drive would work well for handling all the data (models, worlds, smells, taste, sound bites, the program code...) and one was announced a while back (few months?) that was 62.5 TB yet as small as a postage stamp. That doesn't mean it is cheap though.

 

Optical cables have quite the data throughput. I do not know if they could support a MMO though. Another question to ask is how would your data usage affect the entire network? Phone networks limit data due to a lack of infrastructure (and money... lots of money). You and a bunch of people getting upgraded networks and then streaming MMOs 24/7 might not fare well. Of course, this isn't a concern per say for us.

 

The solution to body-jacking is just to only put stimulation technology where needed and only put EEG (or other similar tech) at the motor cortex.

 

--@Kayaba

[CraigD]

Welcome to hypography, Josua! :) Please feel free to start a topic in the introductions forum to tell us something about yourself.

 

A general comment on your post: I think you should avoid paying too much attention to the NerveGear and other devices described in Sword Art Online and other cyberpunk fiction, because ... well ... they’re fictional, and not intended to suggest in detail devices that are actually possible. When Reki Kawahara writes that the NerveGear read and writes to the brain using “high-density microwave transceivers” or that “30% of the NerveGear's weight is from its internal battery”, it’s just for literary effect, to make the story feel more realistic, not a technical suggestion to somebody trying to make a actual video game interface.

 

A specific comment:

Rendering would be a monster to deal with, aside from the CPU. Rendering the level of realism offered by the NerveGear is practically near-impossible with current generation GPUs. Because of this, the GPU needs not only to be compact & power-saving, it must also be powerful enough to render that much pixels.

I think you overestimate the difficulty of rendering pictures realistically enough that they are indistinguishable from reality.

 

A single high end GPU like an Nvidia GTX-1080 can rendering 33,177,600 12-bit pixels (8x UHTV resolution) at 60 frames/s. Test show that this resolution is nearly beyond the limit of human perception to distinguish from higher resolution, except for the detection of aliasing effects (like the “screen door” effect) due to the arrangement of the pixels on the screen.

 

The human retina has about 120,000,000 rod cells, which cannot distinguish the color, and 7,000,000 cone cells, which can. The refractory period is about 0.004 s (analogous to a frame rate of 250 f/s).

 

So, ignoring that retinal cells are effectively 1-bit detectors – that is, they fire or don’t fire when stimulate to their threshold –a single present day GPUs are within a factor of 16 of rendering more data than the human eye can possibly perceive. It’s common place to use many GPUs in parallel (for example, ORNL’s Titan uses 261,632). So I think we can reasonable say that were it possible to interface with every neuron in the human retina, we would not lack for computer processing power to stimulate them as quickly as they are capable of being stimulated.

 

It's said that it was due to the microwave transceivers that redirected all brain signals from the brain to the game with the device. However, as we all know, microwaves don't do something like that in our reality.

They are a form of electricity that could be used to stimulate the brain. It would be horrible and something such as Infrared is much better, but they can do what was described.

I think Josua’s statement is accurate. The “something” to which I think he’s referring, taken from Kawahara’s light novels, is

“The NerveGear's high-density microwave transceivers are capable of accessing the user's brain, allowing it to send fake signals to the five senses of the user. The transceivers are not only capable of inducing fake sensory signals but can also block every movement command from the brain to the body, preventing the player from moving their body while in FullDive to avoid injuries”

Unless your brain cells have had been ontogenetically engineered to be receptive to it (in the visible 4.3 x 10¹⁴ to 7.5 x 10¹⁴ Hz, range), individual neurons can’t be measured or stimulated using EM radiation of any kind.

 

Microwaves are not a form of electricity – a flow of electrons in a conductor – but EM radiation in the 3 x 10⁸ to and 3¹¹ Hz range. Infrared EM radiation is in the 3 x 10¹¹ to 4.3 x 10¹⁴ range.

 

The genes available to ontogenetic to make nerve cells sensitive to and emit EM radiation are limited to those that sense and emit visible light, and lower-frequency EM radiation has proportionally lower resolution, so I doubt EM radiation lower than the visible range can be used to sense or stimulate individual neurons.

 

So, for the NerveGear to work anything like described in its fiction, its “microwave transceivers” would need to be replaced with “visible light transceivers”, and the user’s scalp, skull, and selective brain sections made transparent to visible light. Presently, the only known way to accomplish the first two of these is to remove the troublesome section of scalp and skull and replace them with something transparent, like a glass window.

 

If you give up on the requirement not to stick stuff through your skin and skull, and no messing with your somatic DNA, you can do optogenetics much more comfortably by implanted fiber-optic fibers in your brain. This is far from the neat, non-intrusive consumer product depicted in SAO, but at least it’s physically possible.

[BrainJackers]

Where did you get that number of how many pixels can be rendered? Are you talking about the highest possible? Once throwing in effects and GPU calculations, that will decrease severely, no?

 

Thanks for the correction on Microwave transmitters. 

 

--@Kayaba

[SrOswald]

Ok so real talk, besides all the code and actually getting it to work, how would we read the brain signals transfer it back into readable code while also canceling out the signals the brain is sending out so you move in the simulation but not in real life?

 

Ignoring the amount of ram and video you need because you can just hook it up to a PC like in the actual anime, how would we get the video/ how would it be possible to not look at a screen while also feeling/ everything looking like it's real life?

[BrainJackers]

I thought this was all real talk :P

 

EEG can transfer the brain signals. Ideally, you read the brain itself and cancel the signals out at the spinal cord level. You could also just use a really strong paralysis agent.

 

They don't transfer into readable code but numeric values.  

 

Just because the headset doesn't need a bunch of RAM itself, the PC might. I actually think this is a topic that is quite interesting. How should resources be divided and utilized between the 2 devices? Should it just be a headset?

 

You stimulate the sensory nerves/processing units for those nerves. The earlier you can send the data, the better, as it will be in a less complex format and less specialized (I can get an eye transplant but can't transplant the part of my brain that handles that ocular data. I know this isn't the only issue. I'm just saying...) Look up Transcranial Magnetic Stimulation, Electroshock Therapy, Infrared nerve stimulation...

 

--@Kayaba

[SrOswald]

Your right it is but even so how would we design something like this. Like instead of talking about how it works lets talk about what we need to make it work and how to build it.

[BrainJackers]

You can't say how to build something if you don't know how it works (unless you rely on luck and build a potentially infinite amount of machines).

 

Read

Stimulate

Paralyze

Process

 

Those can be labeled as the main four parts. Reading can be done with EEG. We need away to get a high level of density though. Really high. Beyond that, you don't need to map the whole brain. Just the motor cortex.

 

Stimulation has a few things that can be used. Transcranial Magnetic Stimulation, Infrared, Nanobots, Nanofibers, raw electricity with some method of guidance. There's 5 places to stimulate. Where you process vision, where you process audio, where you process touch, where you process smells, and where you process taste. I didn't list specific places for two reasons. One, I forgot the non-generic labels. Two, there's a couple of places you could process vision (and other senses). Your optical nerves (which are quite standard among human, as demonstrated by the ability of eye transplants) or wherever that data gets formatted (I truly forget the name of that part).

 

Making someone paralyzed also has a few options. You could just make someone emit enough of the sleep paralysis chemical where they can't move (best option, IMO), you could overload their muscles (like a taser; highly not recommended), find a way to cut it off at the spinal cord (where you can still read it before paralyzing it; you would leave the head intact though), or just hold someone down.

 

Finally, you need a processor. I am interested in Cell. It has multiple cores, designed for parallel processing and scientific processing, and that's surely going to be needed with EEG data, 5 senses needing calculating plus movement input, and the standard video game things.

 

Graphics will be very tough. I wrote a paper describing using 1080's to generate graphics in perfect parallel while using Doom as a reference point. It got... extreme. I posted the math somewhere on this site. If you can't find it, I'll post it again.

 

The end statement is, time is needed. EEG needs to get smaller. We need to get better with stimulation. CPUs need to speed up/get better designs. Beyond that, I believe you expect us to actually build a headset (which some of these threads claim), yet we're random internet people talking on a forum. How far could that truly go? That's why the group I founded has a rule about no talking about becoming a company :P

 

--@kayaba

[SrOswald]

Exactly but we've been talking about this for a while and we all have different ideas and understandings and if we designed a prototype everyone would be able to chip in and give there thought on how to improve it so I'm designing one right now but am not going to build it because it will most likely fry my brain if I tried to use it and everyone else on this forum could help build upon the design and we could all work together to make it work.

[SrOswald]

Sorry for double posting but I'm going to post my quote unquote prototype in like an hour or so

[BrainJackers]

I would say the majority of the threads here exist to design a prototype. None have had public success.

 

You think you can design a full prototype for an EEG device, TMS (or other alike technology) device, overlaid on each other, and all it's controlling circuity, and a bridge to a computer/on board computer within a few hours? Why do you think I, or anyone else, haven't done that? It's a tough job.

 

Say you did get a working design and it becomes production ready. Who produces it? If I design part of the PCB, and nothing else, and I get no money from anyone who uses this design, I can sue them. Your idea isn't legally sound for getting a production model.

 

It may seem like I'm being overly critical, yet I am saying the truth. If anything, I'm just not presenting it nicely. Sorry about that. I've seen lots of people try/do what you've discussed and fail. I see no need for someone to try the same thing again when we already have results.

 

Note: I have seen one group that's still active now but I can't comment on their progress.

 

--@kayaba

[SrOswald]

Here's my idea what if we used the original design of the nerve gear helmet just for the shits n' giggles for right now.

 

What we have:

There would be a visor in the front to display the time and battery life.

Indicator led's ex: Pow, Wan, and Blk

Cartridge slot of the game, disk or whatever we use so we don't get killed by untrusted game devs

Ethernet cable so you can connect to the internet directly, highest possible speed, there is no chance of wireless interference

Power cord to directly draw power for everything

Battery which acts as auxiliary power source in case nerve gear is unplugged

Diamond semiconductor CPU to run the software used on the nerve gear and direct commands to different components

Microwave transceiver to scan brain signals

 

What we need:

Some type of way to block brain signals as to not move irl

A way to send artificial brain signals from nerve gear to brain to stimulate pain, smell, etc

We also need some type of code that takes brain signals and turns it into game code and game code into artificial brain signals

 

What we need to fix:

The microwave transceiver cannot pass through water and your brain is made of water/ other scanners fry your brain

The diamond semiconductor CPU is out dated and could probably be updated with something more current

 

Anything I missed or better ideas?

[BrainJackers]

A visor for time/battery life and nothing else is wasteful.
LEDs would be useful.

Having a DVD over cartridge doesn't mean devs can't kill you. The machine should only be allowed to write to sensory locations at a low/safe power.

Ethernet jacks would be useful but it means you're using the device next to your router. It should be an option, not mandatory.

A battery is good to handle emergency shut downs. It shouldn't be hours long though.

Diamond CPUs don't exist in practice right now. Also, the fictional NerveGear connected to a PC with a Diamond CPU. The Nerve Gear itself didn't have one.

The Microwave Device only sent signals, IIRC. The read system wasn't detailed. Beyond that, it wouldn't work in real life.

I've discussed paralysis.

I've discussed sending signals.

You forgot to say reading signals but that's been discussed to death so I'm sure it's just a mistake :)

 

That would be the library for coding games for this machine and is arguably the hardest part. The brain isn't a standard thing. It has standard layouts, assuming no abnormalities or mutations, but neurons are formed via experience and usage. 

 

That's not the issue. The issue is Microwaves wouldn't work at all.

How can something that was never in mass use be dated?
 

You missed the actual work. The "NerveGear" is a fictional object, only to be used for inspiration, and never for reference (except for maybe the end-user experience).
 

Other people may want profit. Beyond that, a fully open device will have some company scoop it up, or you risk people making shoddy copies that kill people, and then you are to blame. If we continue like this, we will never get it done. Correct. Individuals might use this knowledge to join groups that are getting this done though.

 

If you're serious, start a group. Make sure it's laid out properly. Now, you have devotion/interest, and that's awesome. I think you could do something.

 

It's good that random people won't stop you. Working together on the forums is actually harmful in my opinion. Things become messy, companies could steal your data, if you actually do stuff and then wipe you to, your device will have many different issues, people will make cheap ones with no safeties that could kill users... forums also don't offer proper organization. 

 

You don't have to be secret. Just have checks in place. A private group for organization, reducing forum spam, tracking users... keep invites open and share any impressive data.

 

Liability, copyright reasons, keeping the group focused, stopping thefts, preparing a possible commercial launch, making sure the project is covered in bad media by media of murderous clones... If you're serious you can acknowledge these issues and consider them.

 

Credit should be given inside the group but externally, credit should be given to the group as a whole.

 

That's what I have to say. Just again. If you're serious and devoted, start a group.

 

--@kayaba

Edited March 11, 2017 by BrainJackers

[SrOswald]

How do I comment on slack?

[Romer]

I think brain control everything in our body including the nervous system and sensory system,which is compulsory to NerveGear. So all we need to do is take control of the brain and we get it all, I heard there are some information about controlling stuff such as PC for latest, they have special head gear for controlling the PC without using any hand. Controlling stuff is just the beginning, perhaps we can upgrade it (and of course, easier to say then done.). I also know that brain send stimulus, a signal like stuff to control our movement. If we can reverse it, like sending mechanical signal to the brain and the brain will send the real signal to the nerve and the sensory system.

 

Now that i think about it, moving body in the virtual without moving it in the reality is the main problem here, i guess we could make brain sending some sort of signal of sensory while... Wait, what about we can send all the brain signal to one station and that station contain some sort of receptor that accept it make us move in there.

 

Like a virtual body require the real body to send signal to it to make it activate.

[BrainJackers]

Taking control of the body sounds sketchy and there is only actually 6 places we need. Wherever you want to grab each of the five senses (stimulation only) and the motor cortex (read only).

 

There are multiple tools to control various forms of computers with no hands. We have everything from the Kinect to Neurosky headsets to the eMotiv Epoc to professional BCIs.

 

Our brains do to send out stimuli to our motor cortices and various regulatory systems. Ideally you want to block them and use your own. Reversing it wouldn't help and would possibly cause harm (data meant for muscles being fed back to the brain, possibly causing a feedback loop where it tries to send it again).

 

To stop the body from moving in real life, forms of sleep paralysis can be used. I don't know of any virtual body that wouldn't require input in this scenario (RNG/AI isn't what we're looking at). Therefore, your statements actually confuse me. We would send all the signals to a computer and the computer would move the virtual body, yet I believe that has been assumed?

 

--@Kayaba