Anonymous
6/12/2025, 3:04:23 PM No.16696060
You could make a flawless robot hand with Meta's EMG wristband if you got a tesla Optimus hand and hooked it up
Replies:
Thread 16696060 - /sci/ [Archived: 902 hours ago]
Anonymous
6/12/2025, 3:54:04 PM No.16696088
>>16696060 (OP)
If only guys like Elon Musk and other groypers could actually start making working prosthesis', what we have now is an old technology relying on forearms muscle movement to one basic function open and close
how easier my life could have been if I had anything but a retarded claw for a hand, but alas
If only guys like Elon Musk and other groypers could actually start making working prosthesis', what we have now is an old technology relying on forearms muscle movement to one basic function open and close
how easier my life could have been if I had anything but a retarded claw for a hand, but alas
Replies:
Anonymous
6/12/2025, 5:55:59 PM No.16696190
>>16696088
Exactly what I'm saying. The technology is there.
Exactly what I'm saying. The technology is there.
Replies:
Anonymous
6/12/2025, 5:57:00 PM No.16696193
>>16696060 (OP)
In this image, the arm should be a human arm. Sorry about that.
In this image, the arm should be a human arm. Sorry about that.
Anonymous
6/12/2025, 7:45:35 PM No.16696305
>>16696060 (OP)
Anon, what the fuck is putting an EMG on a robot gonna do? If you mean controlling a robot hand with a human hand, it might help a little. Seems like the wrist EMG can figur out where the fingers are. That's nice, but doesn't get you forces on the joints. It doesn't tell you where the object is and doesn't give you tactile feedback. And sure it can inform learning, but until we get westworld tech robot hand dynamics are gonna be different than human hand dynamics. So you've gotta figure out how to map to the different dynamics. But honestly, learning from people is fucking dumb. Robots should be able to learn on their own. If you have to get a trillion hours of humans wearing fancy gloves, that doesn't scale.
>flawless
Optimus hand isn't SOTA. It's tactile sensing is utter shit. Resolution's too low. Only dumb computer scientists think you can do manipulation only with vision
Anon, what the fuck is putting an EMG on a robot gonna do? If you mean controlling a robot hand with a human hand, it might help a little. Seems like the wrist EMG can figur out where the fingers are. That's nice, but doesn't get you forces on the joints. It doesn't tell you where the object is and doesn't give you tactile feedback. And sure it can inform learning, but until we get westworld tech robot hand dynamics are gonna be different than human hand dynamics. So you've gotta figure out how to map to the different dynamics. But honestly, learning from people is fucking dumb. Robots should be able to learn on their own. If you have to get a trillion hours of humans wearing fancy gloves, that doesn't scale.
>flawless
Optimus hand isn't SOTA. It's tactile sensing is utter shit. Resolution's too low. Only dumb computer scientists think you can do manipulation only with vision
Replies:
Anonymous
6/14/2025, 7:01:42 PM No.16697748
>>16696305
My idea is to give an emg wristband to people who have lost their hand in an accident. They would use the emg wristband to control the robot hand.
Please ignore the image. The part below the wristband is supposed to be human.
My idea is to give an emg wristband to people who have lost their hand in an accident. They would use the emg wristband to control the robot hand.
Please ignore the image. The part below the wristband is supposed to be human.
Anonymous
6/14/2025, 7:05:58 PM No.16697750
>>16696305
1:1 Finger tracking is all you need to make a perfect robot hand in terms of input and tesla has the most advanced robot hand in the world.
In my opinion, tactile sensing isn't that important. Most robot hands on the market don't do it.
My point isn't that this will be as good as a human hand, but that it will be by far the best bionic hand in the world. It will be able to make every pose a human hand can make and more. That cannot be said about current bionic hands.
We have the technology. Why it isn't being done is beyond me.
1:1 Finger tracking is all you need to make a perfect robot hand in terms of input and tesla has the most advanced robot hand in the world.
In my opinion, tactile sensing isn't that important. Most robot hands on the market don't do it.
My point isn't that this will be as good as a human hand, but that it will be by far the best bionic hand in the world. It will be able to make every pose a human hand can make and more. That cannot be said about current bionic hands.
We have the technology. Why it isn't being done is beyond me.
Replies:
Anonymous
6/14/2025, 7:07:28 PM No.16697753
>>16697750
In my opinion, the next technological revolution after artificial intelligence will be the bionic revolution
In my opinion, the next technological revolution after artificial intelligence will be the bionic revolution
Anonymous
6/14/2025, 7:08:42 PM No.16697756
>>16696190
>The technology is there.
has been for years, it's just that it's expensive as fuck or at the very least they want to pass it as such, remember a vid about US gov nasa or whatever that might have been making a flawless robotic arm prosthetic that literally did the thing, they gave it to a guy, he was wow this is amazing, then he had to give it back, just an experiment, because lol not for you poorfag, just science things
>The technology is there.
has been for years, it's just that it's expensive as fuck or at the very least they want to pass it as such, remember a vid about US gov nasa or whatever that might have been making a flawless robotic arm prosthetic that literally did the thing, they gave it to a guy, he was wow this is amazing, then he had to give it back, just an experiment, because lol not for you poorfag, just science things
Replies:
Anonymous
6/14/2025, 7:40:14 PM No.16697778
>>16697756
I think I know the video you are talking about and I am saying it will be better than that arm, although it will only work for people who have lost down to the wrist, which is still quite a lot. It would also be used by non disabled people just for fun haha. Everyone could have one.
As for cost, I do not expect it to exceed $100,000. Meta's wristband is pretty cheap to manufacture and Tesla Optimus will eventually go into mass production, costing $30,000 for the whole robot so imagine how cheap just a hand could be. Apart from that, you would need a central computer to allow the two pieces of tech to communicate, but computers are pretty cheap these days.
There is no reason not to do it other than sheer negligence.
I think I know the video you are talking about and I am saying it will be better than that arm, although it will only work for people who have lost down to the wrist, which is still quite a lot. It would also be used by non disabled people just for fun haha. Everyone could have one.
As for cost, I do not expect it to exceed $100,000. Meta's wristband is pretty cheap to manufacture and Tesla Optimus will eventually go into mass production, costing $30,000 for the whole robot so imagine how cheap just a hand could be. Apart from that, you would need a central computer to allow the two pieces of tech to communicate, but computers are pretty cheap these days.
There is no reason not to do it other than sheer negligence.
Replies:
sage
6/14/2025, 7:44:34 PM No.16697783
Replies:
Anonymous
6/14/2025, 7:56:00 PM No.16697792
>>16697783
Why?
Why?
Anonymous
6/14/2025, 11:36:33 PM No.16697933
>>16697783
You are just too stupid to understand what I am saying.
I'm saying that a handless human arm could control a robot hand with an emg wristband. The movement would be 1:1. It would feel like having a hand again. It would change the world.
You are just too stupid to understand what I am saying.
I'm saying that a handless human arm could control a robot hand with an emg wristband. The movement would be 1:1. It would feel like having a hand again. It would change the world.
Replies:
Anonymous
6/15/2025, 7:58:14 PM No.16698607
>>16697933
by 1:1 I mean perfect finger tracking. This is huge! Why is no one doing this?
by 1:1 I mean perfect finger tracking. This is huge! Why is no one doing this?
Anonymous
6/17/2025, 5:42:07 AM No.16699744
Replies:
Anonymous
6/17/2025, 12:27:36 PM No.16699871
>>16699744
Yeah probably. My one would not be hydrogen powered though
Yeah probably. My one would not be hydrogen powered though
Anonymous
6/17/2025, 1:55:46 PM No.16699907
>>16696060 (OP)
Do people actually think like this or are you memeing?
Do people actually think like this or are you memeing?
Replies:
Anonymous
6/17/2025, 1:57:00 PM No.16699908
>>16699907
Say why it wouldn't work. It's called not being an npc you should try it
Say why it wouldn't work. It's called not being an npc you should try it
Replies:
Anonymous
6/17/2025, 1:58:33 PM No.16699910
>>16699908
Don't think I can put it into words that you'll understand.
Don't think I can put it into words that you'll understand.
Replies:
Anonymous
6/17/2025, 2:01:25 PM No.16699912
>>16699910
Do it anyway. Surprise me. Make my day.
Do it anyway. Surprise me. Make my day.
Replies:
Anonymous
6/17/2025, 2:54:47 PM No.16699949
>>16699912
Oh wait you can't because you have no argument. Check mate.
Oh wait you can't because you have no argument. Check mate.
Anonymous
6/17/2025, 6:30:09 PM No.16700105
>>16697750
You have outed yourself as a dumb computer scientist
>tactile sensing isn't important
Video related https://www.youtube.com/watch?v=HH6QD0MgqDQ
Including tactile sensing improves performance
https://arxiv.org/abs/2504.13618
In order for robots to be useful they need to be very reliable. A home robot that breaks a plate every month would be thrown in the garbage. Getting 99% reliability and higher will probably require tactile sensing. Video and finger position alone can't tell you if a held object is slipping. In the real world friction varies. You can't see friction, you can make guesses, but these are only guesses. Lower than expected friction means the object can start slipping and something must be done to stop it. Video isn't the best for sensing slip, the object might move less than a millimeter.
>most robot hands don't do
Because they're driven by hype rather than applications. Hands that look like human hands seem more impressive, but they're currently a terrible form factor for high resolution tactile sensing. Tell me, what exactly are these hands doing? Are they doing anything of value rather than just rough grasping?
Tactile sensing enables a simple claw gripper to manipulate flexible cables
https://gelsight.csail.mit.edu/cable/
>pose
Copying poses is great if you just want to make robots that look nice, but don't actually do anything like animatronics and toys. Copying poses is useless if the robot isn't smart enough to use the hand or the hand doesn't provide enough information to carry out useful tasks.
You have outed yourself as a dumb computer scientist
>tactile sensing isn't important
Video related https://www.youtube.com/watch?v=HH6QD0MgqDQ
Including tactile sensing improves performance
https://arxiv.org/abs/2504.13618
In order for robots to be useful they need to be very reliable. A home robot that breaks a plate every month would be thrown in the garbage. Getting 99% reliability and higher will probably require tactile sensing. Video and finger position alone can't tell you if a held object is slipping. In the real world friction varies. You can't see friction, you can make guesses, but these are only guesses. Lower than expected friction means the object can start slipping and something must be done to stop it. Video isn't the best for sensing slip, the object might move less than a millimeter.
>most robot hands don't do
Because they're driven by hype rather than applications. Hands that look like human hands seem more impressive, but they're currently a terrible form factor for high resolution tactile sensing. Tell me, what exactly are these hands doing? Are they doing anything of value rather than just rough grasping?
Tactile sensing enables a simple claw gripper to manipulate flexible cables
https://gelsight.csail.mit.edu/cable/
>pose
Copying poses is great if you just want to make robots that look nice, but don't actually do anything like animatronics and toys. Copying poses is useless if the robot isn't smart enough to use the hand or the hand doesn't provide enough information to carry out useful tasks.
Replies:
Anonymous
6/17/2025, 6:31:17 PM No.16700107
>>16697750
>finger tracking is all you need
Great, so in order to train a robot to use its hands you need 1 trillion hours of humans wearing specialized gloves doing every task imaginable. The glove are expensive, the people are expensive, and the robot isn't guaranteed to reach human level performance. The training data might not cover everything it needs to do in the real world or the robot just doesn't learn anything from the data for whatever reason. Relying on humans to train robots doesn't scale.
>finger tracking is all you need
Great, so in order to train a robot to use its hands you need 1 trillion hours of humans wearing specialized gloves doing every task imaginable. The glove are expensive, the people are expensive, and the robot isn't guaranteed to reach human level performance. The training data might not cover everything it needs to do in the real world or the robot just doesn't learn anything from the data for whatever reason. Relying on humans to train robots doesn't scale.
Replies:
Anonymous
6/18/2025, 12:38:35 PM No.16700626
Anonymous
6/18/2025, 12:41:11 PM No.16700628
>>16700105
it would still be better than anything that exists in the world today. Obviously tactility is an improvement.
it would still be better than anything that exists in the world today. Obviously tactility is an improvement.
Anonymous
6/18/2025, 12:42:35 PM No.16700629
>>16700105
I am not suggesting the robot hand be used by a robot, I am suggesting it would be used by a human. Some, who have lost their hand in an accident, and some who haven't and just want to use it for fun
I am not suggesting the robot hand be used by a robot, I am suggesting it would be used by a human. Some, who have lost their hand in an accident, and some who haven't and just want to use it for fun
Replies:
Anonymous
6/18/2025, 12:45:13 PM No.16700630
if you can jerk off with it without crushing your junk it should be good...
Anonymous
6/18/2025, 10:19:11 PM No.16700995
>>16700629
EMG controlled prosthetic hands already exist. Facebook's sensor may not work with a stump. Just use Neuralink or GTFO for control. Tesla's hand has the actuators in the forearm, it's not very compatible with a stump.
EMG controlled prosthetic hands already exist. Facebook's sensor may not work with a stump. Just use Neuralink or GTFO for control. Tesla's hand has the actuators in the forearm, it's not very compatible with a stump.
Replies:
Anonymous
6/19/2025, 3:26:00 PM No.16702007
>>16700995
I know but I'm saying that the two best in the world examples of the technology could be used together
I know but I'm saying that the two best in the world examples of the technology could be used together
Anonymous
6/19/2025, 3:27:02 PM No.16702010
>>16700995
I know they already exist but there has never been a good one
I will admit it will be harder to control with a stump than by someone who still has their hand, but with training it can be done
I know they already exist but there has never been a good one
I will admit it will be harder to control with a stump than by someone who still has their hand, but with training it can be done
Anonymous
6/19/2025, 3:30:22 PM No.16702014
>>16700995
meta's wristband wouldn't take surgery
meta's wristband wouldn't take surgery
Anonymous
6/20/2025, 11:19:54 AM No.16703182
>>16696060 (OP)
Anyway, it will be done by hobbyists once both products are sold to the public so it is a matter of time.
Anyway, it will be done by hobbyists once both products are sold to the public so it is a matter of time.
Anonymous
6/21/2025, 11:50:21 AM No.16704094
https://youtu.be/UgC7la_3IIA?si=PJrIz6lFPM22DV8G&t=414
Replies:
Anonymous
6/21/2025, 1:09:08 PM No.16704152
>>16704094
Trash. The one I am suggesting will be bar none the best in the world
Trash. The one I am suggesting will be bar none the best in the world
Replies:
Anonymous
6/21/2025, 1:14:39 PM No.16704159
>>16704152
Given most of the Tesla hands equipment is in the arm, I would say what I posted is more realistic for a human.
https://www.youtube.com/watch?v=qgAw5eU9sjY
Given most of the Tesla hands equipment is in the arm, I would say what I posted is more realistic for a human.
https://www.youtube.com/watch?v=qgAw5eU9sjY
Replies:
Anonymous
6/21/2025, 1:36:53 PM No.16704177
>>16704159
Use cases, as well as being for amputees would be remote teleoperation and training for autonomous hand movement and many more use cases beyond that. Would not matter. Meta is missing something huge by not doing this. They are using their groundbreaking tech for fucking glasses.
Use cases, as well as being for amputees would be remote teleoperation and training for autonomous hand movement and many more use cases beyond that. Would not matter. Meta is missing something huge by not doing this. They are using their groundbreaking tech for fucking glasses.
Anonymous
6/21/2025, 1:44:00 PM No.16704183
>>16704159
gen 3 hand would be used
gen 3 hand would be used