Device for general purpose augmentation - "ACUMA"? (GPT-3 invented the name)

Projects
81

That’s fine.
Even if keeps me sharing the same suspicion of @Eriequiet … XD

I think the main bit we are confused here is…

  • You got a concept: :white_check_mark:
  • You have some ideas of what it could do: :white_check_mark:
  • You heard about some technologies you could use: :white_check_mark:

Yet… Having an Idea is not the same thing as having a project.

This isn’t to do with how complex it is… a Project can be of any complexity and ambition.
The main differential here is that a project has direction and reasoning.

While an Idea can be “oh, what if we did something with magnets on the tracks to get a train floating above it to get a frictionless transport?”

the equivalent project would be much more focused on what is that something? Why we think it would work in a larger scale? What exactly is the type of magnets to use, and why? Etc,etc…

And please don’t get us wrong.
This is not me trying to be “elitist” and “demanding an academic level research”…
Far from that!

It’s important to understand if you need help with coming up with a project or if you need help with bringing a project to life, so we can help you better:

If you’re still in the Idea development stage of your concept, we can help you with new ideas, offering advice, reviewing your choices, etc…

But while you’re asking for help with steps that should only happen after the project stage, we need to “see” that project.
Again, not being elitist, but… how on earth can I make you a 3D model of the printable parts for the gloves if you haven’t showed me how they are meant to look/work?

Hope this makes it clear what is needed before you can get any help, and why…

82

Well tbh idk exact what youd have to do to get this to work but for the gloves I was thinking using those LucidVR gloves but modifying them so the hardware is still functional but father up on the arm (since it’s a suit) and stretched out kinda somehow so it’s not as bulky (since normally you have somewhat bulky gloves), that way the gloves could be worn while doing normal everyday activities and not get in the way

1 Like
83

I’m happy to keep bouncing Ideas, but ultimately figuring that is exactly what will turn this idea into a project.

That said:

I see your reasoning there. It would integrate it better and make it less bulky.

Unfortunately that would not work with rotary-decoder-based gloves without some major overhaul on how they work.

a Rotary based decoder is nothing more than a spring loaded spool of wire (or something similar) with a sensor to detect how much of the wire is out of the spool.
With this you have one numeric value which you can correlate to something…

In the case of the LucidVR gloves, each spool gives you a number that trannslates to the state of one finger, like this:

Measure how much of the wire is out of the spool when you stretch your finger. This is your baseline
Then measure how much of the wire is out of the spool when you close your finger. This is your maximum value.

Then you get the software to extrapolate a 0-1 value from “Current length out” - “the baseline value”, and then folds the finger in the 3D model accordingly.

There are numerous issues with that approach. One of them is that you only track one value for each finger, so your hardware cannot tell when you are spreading your fingers sideways.
Worse, if you spread the fingers sideways it will think you are folding the fingers lightly instead.

Why am I explaining all that?

Because once you move the spools up to the arms, the wires will pass through other joints. at least the wrists.
When this happens, if you bend your wrist even on the slightest, the system will think you are bending all your fingers instead.

So… for every joint you go through, you would need to add some extra spools and wires to track said joint, just so you can compensate for the joints further ahead…

Not only that would make for a system with a huge margin of error, but you would also end up with about 20 spools attached to your shoulders with wires running on the outside of your arms… making it so you must be standing up and it would easily be caught on anything that touches your arm…

Not to mention that the haptic feedback provided by that system would be a huge mess and you might feel your elbow being pulled back because your finger touched a 3D object.

Also, increasing the scale like that would also put some strain on the bluetooth module needed, on the batteries, and a few more technical issues…

In short… LucidVR is a system which barely works as finger tracking, and you for sure don’t want that style of sensors on a suit.

There are many other methods for tracking motion. Some of which are even mentioned on the video from the link you posted.

You just need to read into a bunch of them, try to figure out what could work and what would cause issues, and then try to come up with a theory to fix those issues.

For this stage you don’t need neither money nor cad skills nor parts. it’s purely reading and thinking.

If it serves as an incentive, if you can crack how to effectively make up cheap and efficient full body haptic suits, you’re done for life!
That’s an easy grab for investment money, and currently full haptic suits selling to military personal go for a few hundred thousand a pop… and that’s still on experimental stages.

So I can see that suit of yours, with a revamped look, being a goldmine.
If you can crack it!

84

Well I don’t know if this is ridiculous or not but… Could infrared LEDs and cameras be used somehow? Or would sunlight interfere? Or possibly using the normal design but kinda making the fingers of the glove extended slightly to kind of make it easier to use (once you get used to your fingers being like that, as it’d actually allow you to add sensors to the extensions now that I think of it. That’s a neat idea. Especially since the gloves have feedback so could feel electromagnetic fields and stuff with this. And maybe put lasers in the extensions bc why not have lasers (maybe blue lasers idk but that’d be sorta ridiculous)

85

Yep. they could!

With modern computervision software you can even achieve a pretty good result with no infrared markers.

I’ve done a small motion capture project back in 2015 using only some old PlayStation mocap cameras we got on ebay for about 10$.

Trick with those is… they are set up outside of the body so they can capture all your movements. So not exactly useful for a suit.

Another option is to go down the LeapMotion route:
Get a camera on the VR visor or on your chest, and then that captures the gestures you do in front of it.

That can be done with a cheap arduino and camera.
But training the computervision algorithm can easily take 6 months to a couple of years.

What would be the “normal” design?

86

Normal LucidVR gloves design. But with casing (3d printed) protecting the electronics

87

these glooves have the issues of:

  • being unreliable
  • making it so manipulating objects becomes harder
  • being clunky
  • relying on wires that can catch up on stuff
  • being a bad choice for tracking other joints

By making the fingertips longer, we effectively solve none of these issues, but add a new one:

  • manipulation becomes even more limited.

The electronics being exposed is not an issue.
They are only exposed while it’s a prototype.

88

What do you think would work for it then? Like to fix the issue? Would maybe the types of encoders used in old mechanical computer mice work?

89

I personally don’t know.

I think that spool-operated systems would have too many issues to be worth fixing. but you might find a way to overcome them.

There are flex/bend sensors, which are amongst the cheapest types of sensors for this use case I can think of right now, but even those would make your project considerably expensive.

Cameras + computer vision are amongst the cheapest options I can think of, but that requires a lot of coding work from your side.

The thing is… you’re trying to do something very similar to projects that quite some people are trying to do: haptic/self-contained-mocap suits.

So if there were an easy answer that I could just say “do this”, then it would already be on the market!

This doesn’t mean you can’t do that.
Just that there isn’t an answer out there.

You need to study your alternatives and come up with theories.
Once you have well funded theories I bet a lot more people here would gladly offer their opinions as well!

for example, in this question:

You’re not making clear which old mechanical mice you refer to. there are at least 3 old technologies involved there.

And then, even if you specify “would maybe something like those old mice with a ball in it work?”… I would still have no clue how would that work out.
What would be done with the balls? or where/how would you place the rolling bits to detect movement?

you need to give me thought up theories of how you plan on achieving that so I can bounce back any useful thoughts to you.

90

I’m referring to like the small sensors in those mice (with a ball)

91

:point_up_2:

92

I’m just referring to the sensors for rotation. From what I understand, some of those mice have like 2 small rotary encoders (possibly using LEDs and small discs with slots in them in some circumstances). Smaller sensors=less bulk

93

That’s true.

But then, how would you make use of those sensors?

In a trackball mouse they are all trapped around the ball being pressed against it, so that when the ball moves, friction rolls them.

But in a suit…?

94

Well I did a search and found this mechanical mouse rotational encoders at DuckDuckGo

Obviously they’re small so it shouldn’t be too difficult too add stuff to make them work for this and not have them be too big

95

but how you plan on using them?

96

Basically that’d be the rotational encoders

97

that’s what those parts are.

But how would you use them?

It’s like me saying: “I could use elastics”.
Yes, I could… but how would I extract computational data that correlates to each finger being tensioned, from “using elastics”?

98

Also this may help too https://www.instructables.com/Somatic-Data-Glove-for-the-Real-World/ adding more sensors and haptic feedback would make it very similar to LucidVR (also making 2 I don’t think this person made one for each hand). At least how I’d be planning to use it

99

Do you think that may work? Since that actually is designed to control a wearable computer (I believe the person who made it had a modified Epson Moverio BT-200, I saw this in a YouTube video).
I’d just remembered seeing this so I looked it up. I thought maybe this design would be easier to use for this. Do you think that’d work?

100

At least as a simpler version of the glove? Possibly adding more sensors later on for like proper 3d tracking and stuff but still.