For those curious, Tensor-Board graphs for the final trained model showed "block frequency" way higher than "punch frequency". For some reason the AI resorts to blocking more than punching. I will be sharing more data and answering questions regarding the simulation over on Twitter: x.com/cozmouzz
If they are trying to respond to each other's movements also, from the total begining, even before they learned how to stand, then you might be pushing them way too hard.
I would say that you need to keep an energy parameter. The reason we don’t see fighters constantly using these dance movements is because it would drain their energy. Add a limited energy parameter and you see better results.
@@philv2529Yeah and nobody in MMA fights like they are doing Capoeira... Besides, Capoeira is relatively slow, sweeping movements, not hectically spazzing out in random patterns. A human could do this for maybe 20 seconds before being exhausted
@@RRKSreward is a drug for them, that's why they got kinda worse, you need to make penalty for existing so they get dads belt every nanosecond so it would be ultra instinct instead of drunk uncles fighting each other
at the end of the day, thats amazing, its a energized rock that thinks by itself (the computer), and they're evoluting as our species did. if you let them train billions or trillions of fights, they gonna come with insane techniques i guess.
@@nanja7773 well unlike bot comments the message actually makes sense as a reply to the original comment and wasn’t just ‘this is the clip you’ve all been looking for’ or whatever
@@rumplstiltztinkerstein One thing is standing up, another is to balance. there's a huge mechanical difference. You can observe it in babies. They first learn to stand and then to walk.
It might look dumb at first glance but they are feinting each other faster than a human could almost instantly leaving no openings so they have to constantly change their angle of attack and cancel it/turn it into a feint midway through, maybe you could get better results if you train 1 AI to be proficient at attacking an evasive target that could also block (likely with motion capture), another one to be defensive by evading and blocking, and then combine both. Or maybe I'm just wrong and it will be the same result.
They aren't feinting each other because they weren't trained to feint, there are multiple situations where a normal hook would get them but they don't throw it cause they don't know it. When you are about to throw a punch but the person dodges you don't need to cancel the entire punch, you just adjust to where the person is or you throw another type of punch
They are not doing shit, they are figuring out how to punch. We know how to punch because we have innate information about the functioning of our bodies. They are not even aware of their bodies. They just move a certain combinations of joint around and if it satisfies some parameters they tend to repeat It. I think you should just leave them training for months and we will see progress.
I think the reason it looks jittery is that they have the minimum possible reaction time. Meaning that they would react to the opponent's movement as soon as it happens. And if both avatars act like that, it would explain why the block frequency is higher than the punch frequency.. because both of them are reacting very quickly to any movements the opponent makes. Also if you have this in mind while watching the fight, you'll realise how cool these combat robots are, and that humans probably wouldn't be able to fight them in the ring
It takes the human 160 milliseconds to register an opponents movement. The average punch travels 31.68 feet in that amount of time (starting at 0 mph and ending at 12 mph at impact). Part of fighting is learning to "read" the opponent, typically called "fight IQ," by experienced fighters. That's the part that AI will need in order to conquer mankind.
you have to keep in mind that they have very limited experience and its incredibly difficult for them to learn from experiences with a real human fighter (or hell, even against each other).
8:26 i know what its like to be the gold guy in that scenario. Ive fought silver guy outside a club and just stood there like 'wtf are you doing bro?' While the guy took himself out 😂😂😂
Looking at the ragdoll model, it has a lot of "joints" for its hands but only two for the feet, our feet are basically hands as well, so it might help with learning to balance if you made the feet more complex as well. Also, we are able to use our muscles with unequal emphasis stabilizing 0-100% on one leg alone, maybe there could be a mass X for the upper body which needs to be stabilized but can be shared unequaly between the legs, with shifting percentages. Anyhow, they need way more standing and walking before they can fight. :)
What everyone else said. It is hilarious, and it is because there is no need to be economical with an energy reserve. Some cumulative refractory period on the muscles may help. It should learn to not exhaust them, so be static stable instead of dynamic stable.
This gives an idea for a fight between 2 robots with ai operating them, with knowledge of all sorts of martial arts.. imagine one robot just uses a 540 kick to swiftly end the round.
A sentient being that you use as a slave, started to train for fighting but to your eyes is just a dance. POV: you're a portuguese patron in 17th century Brazil
Get one of these bots that do know how to fight and put it against the new bot that doesn’t know how to fight. Keep the simulation running until the new bot wins.
Was watching UFC the other day and briefly wondered what it would look like if you trained an AI to fight, and here it is! Guess Google was reading my mind again or something
I think a big part of the problem with the final product here is that during balance training, the bots learned to use their arms as integral parts of their balance; the punch training was awkward afterwards because they were still trying to use their hands to balance too much to really ever extend them meaningfully far from their own bodies. I don't necessarily know what the solution is - balance is just kind of a tricky thing to train- but some way to eliminate or reduce the allowed use of arms during balance training would probably allow them to throw punches more freely in later stages
Changing the training so that it rewards keeping their torso as close as possible to vertical could also help. And rather than random kick force to the hips, adding random forward momentum to the hands (and maybe torque to the hips, for twisting into the punch) could help them learn to compensate for punches, which could translate more effectively to the punch training stage
like what others have said. an "energy" cost per movement and force would be helpful, as it limits the movement necessary to keep balance, and also avoid erratic motions. I would also like to add that where a person is hit could affect their balance, especially the head, as that is where our sense of balance is being controlled. we can also observe and learn from our opponent's movement as well, and adapt accordingly. you could train the AI to learn from different fighting styles, give them limited energy, a sense of balance and movements to control it, and have them react to getting hit, or react to movements that would get them hit, sort of like an eye tracking/ object proximity detection. these are highly advanced and complex to simulate. but considering these factors would make you understand that there's more to fighting than movement alone
The agent needs a response latency limit both in the perception and the reaction phase. As others stated, it also needs an energy cost that scales with movement size and force. Small movements like balancing or barely moving the arm to feint should cost almost nothing, large movements should cost a little, and consecutive high-effort movements should cost the most and add up to a diminished energy recovery over time. This energy system would be the single most complex part to program correctly, but if done right, it should result in much more realistic movements once you get the training balance right between energy conservation and pacifism. *EDIT:* Also, their reaction strengths need to be imperfect. Lots of times fighters can overcompensate for feints, which then allows their opponent to capitalize with an actual attack, or sometimes even anticipate actions that never come, which usually ends up looking like erratic behavior to the opponent. But these anticipatory moves can also be used to lure the opponent into attacking, which can then allow a fighter to counter-capitalize on an opening that their fake reaction created.
You should start by making them play tag. Take force out of the equation. 1 point for touching their forearm, and 3 points for touching their head, and 5 points for the chin. They should (at the very least) learn how to manipulate the opponents guard and create opportunities to land clean jabs.
Idea: train the AI in a simulation, then make a robot irl with the exact same properties as the ai in the simulation and use your graphics card to make the ai behave the same way irl
Maybe reward it for taking 1 step a second that way it learns to step and fight as importance number 1. Once it takes 20 steps in 20 seconds then it knows that footwork is important as hell. But if you reward it more than once each second it might start tap dancing for needless points. Then teach it if it KOs its enemy it gets x10,000 points so the AI also tries to finish the fight instead of bouncing around for extra points.
Do you also simulate exhaustion effects? I am pretty sure the result would be different. Why it is beneficial now? I would guess because it comes at no cost and the opponent is unable to 'read' your moves resulting in less hits received.
Some mentioned energy would help, I agree. Others said they're reacting to each other too fast. I would add to that, are they given a reflex window and response time for their actions? This could play into the energy system where they are given a response time, then allowed to speed it up or slow it down at the cost or conservation of energy drain, respectively. This would motivate their training towards playing slower and smarter. Additionally, to motivate aggression, penalize them for failing to strike, or getting hit. I believe this, coupled with a little more balance training that also uses energy and rewards an upright posture, would turn these agents into boxing legends.
Would be pretty cool to see it learn different fighting styles from boxers. Imagine being able to put them in vr games so you could "fight" prime Mike Tyson, Muhammad Ali, Mayweather, pacman etc
i think there was something missing from this: the human body wont really be hurt much at all by the very soft punches that the AIs are throwing, so i think there needed to be a minimum cap of force for a punch to count as well. this way they would only get rewarded if they threw a harder punch, which would probably end up looking a lot more "correct" it also seems like they are waiting each other out for one to fall before the other sometimes instead of trying to hit. i think something was missing from the balancing training to make them more stable but im not entirely sure what, though maybe it was multiple different sources of balance to train off of, like each hour of training or something would pick a new random balance source. surprisingly only the one source for a punch seemed to work fine, though, it does look like they adapted that into different kinds of punching like a hook and a jab and an uppercut.
is it possible to train a bot, to chase somthing an atack it? jump maybe? Then add a player controller after its fully trained and see how it will fight you?
It appears that the AI is having a hard time trying to balance itself while mantaining the distance to the opponent. They throw occasional punches only when these 2 variables are OK and the punch won't make them fall...
![BROLY Se Descontrola Y Casi Acaba Con VEGETTA Y GOKU [Español Latino]](http://i.ytimg.com/vi/GC11i7Ul-1Q/mqdefault.jpg)
![BROLY Se Descontrola Y Casi Acaba Con VEGETTA Y GOKU [Español Latino]](/img/tr.png)
For those curious, Tensor-Board graphs for the final trained model showed "block frequency" way higher than "punch frequency". For some reason the AI resorts to blocking more than punching. I will be sharing more data and answering questions regarding the simulation over on Twitter: x.com/cozmouzz
I know how to solve it @cosmouz make the punch rewards a quadratic function in punch velocity, currently yours seems linear
@cozmouz whats the music at 3:54 ??? Can i have the title? Could anyone provide me the title?
If they are trying to respond to each other's movements also, from the total begining, even before they learned how to stand, then you might be pushing them way too hard.
@@bluestone-gamingbg3498 Ben Elson - 89
sir, feed them some basic martial arts data and then challenge them with 5 more simulations. see if they improvise
I would say that you need to keep an energy parameter. The reason we don’t see fighters constantly using these dance movements is because it would drain their energy. Add a limited energy parameter and you see better results.
i see any human doing this dance movement will get tired extremely quickly
any machine will also wear before get into the fight
Yeah, so hopefully they would learn to min max their stamina
But that martial art already does exist it's called capoeira
@@philv2529Yeah and nobody in MMA fights like they are doing Capoeira... Besides, Capoeira is relatively slow, sweeping movements, not hectically spazzing out in random patterns. A human could do this for maybe 20 seconds before being exhausted
yes but they might get tired at the same time
Fights outside the pub 3am in the morning 🥶
After drinking 1Million gallons of alcahol
@@10aelbeg22 Due to the speed, i'd certainly think there's some coke, meth or whatever "up" drug involved as well. Maybe all of them
@@RRKSreward is a drug for them, that's why they got kinda worse, you need to make penalty for existing so they get dads belt every nanosecond so it would be ultra instinct instead of drunk uncles fighting each other
Mfs had hundreds of millions of fights just to end up fighting like my drunk uncle
at the end of the day, thats amazing, its a energized rock that thinks by itself (the computer), and they're evoluting as our species did. if you let them train billions or trillions of fights, they gonna come with insane techniques i guess.
Already looks insane ;)
Nah this was funny 😂 made me laugh out loud
@@gat0anonimo hey! Are you my nephew? xD I'm better sober damnit
Have you considered that your drunk uncle is a synth?
The spastic synthwave dancing is genius
reminds me of ua-cam.com/video/cqPD4A-bKNA/v-deo.html 😄
@@deeOOghI clicked this expecting you to be a bot comment, I was happily surprised, thank you for sharing drunk russian dancers with us
@@roux6715u thought they were a bot and still clicked?
@@nanja7773 well unlike bot comments the message actually makes sense as a reply to the original comment and wasn’t just ‘this is the clip you’ve all been looking for’ or whatever
That's literally the only reason I watched the whole video... then WATCHED IT AGAIN 😂😂😂
i am no longer afraid of AI killing off the human race
You’re fucked in a dance battle
Killing us with their awesome dance moves
The ai watched rumble the movie
@@Anthonybrother Yeah, we're going get served!
Remember a few months ago when AI couldn't draw a hand?
Then Sora happened.
Fights with your brother be like:
they are worse than toddlers 😂
More like telling someone with adhd to stand still
More like your little brother trying to fight you
@@antonellisamuele6569plot twist: he is the little brother
@@rotorblade9508nuh uh
They are trying to balance using their hands not standing on feet
69 👍
True, so do we... That is why we balance our arms as we walk and run.
Yeah gravity needs to be implemented here so that they're More planted
@@RenzitoARG Are you saying that you are incapable of standing up without using your hands for balance?
@@rumplstiltztinkerstein
One thing is standing up, another is to balance. there's a huge mechanical difference.
You can observe it in babies. They first learn to stand and then to walk.
It might look dumb at first glance but they are feinting each other faster than a human could almost instantly leaving no openings so they have to constantly change their angle of attack and cancel it/turn it into a feint midway through, maybe you could get better results if you train 1 AI to be proficient at attacking an evasive target that could also block (likely with motion capture), another one to be defensive by evading and blocking, and then combine both.
Or maybe I'm just wrong and it will be the same result.
No, I read multiple tips in the comments thatcwill for sure aid the models in the end, including yours.
They just need a reaction time. If they reacted a few hundred milliseconds later to each movement of their opponent, they wouldn't do this.
That first thing you said is definitely not happening.
They aren't feinting each other because they weren't trained to feint, there are multiple situations where a normal hook would get them but they don't throw it cause they don't know it. When you are about to throw a punch but the person dodges you don't need to cancel the entire punch, you just adjust to where the person is or you throw another type of punch
They are not doing shit, they are figuring out how to punch. We know how to punch because we have innate information about the functioning of our bodies. They are not even aware of their bodies. They just move a certain combinations of joint around and if it satisfies some parameters they tend to repeat It. I think you should just leave them training for months and we will see progress.
The "damn bees attacking my face" dance
I think the reason it looks jittery is that they have the minimum possible reaction time. Meaning that they would react to the opponent's movement as soon as it happens. And if both avatars act like that, it would explain why the block frequency is higher than the punch frequency.. because both of them are reacting very quickly to any movements the opponent makes.
Also if you have this in mind while watching the fight, you'll realise how cool these combat robots are, and that humans probably wouldn't be able to fight them in the ring
It takes the human 160 milliseconds to register an opponents movement. The average punch travels 31.68 feet in that amount of time (starting at 0 mph and ending at 12 mph at impact). Part of fighting is learning to "read" the opponent, typically called "fight IQ," by experienced fighters. That's the part that AI will need in order to conquer mankind.
1 mean left hook and they're done
The AI just has ultra instinct, that’s why it can react so fast smh my head
you have to keep in mind that they have very limited experience and its incredibly difficult for them to learn from experiences with a real human fighter (or hell, even against each other).
@@Oren_is_tiredwait for them to complete;y dodge then punch you
“Bro fighting demons”
This feels like watching 2 stand users fight but I don't have a stand
This is a banger
Fr
Lmfaoooo
We all been watching 15 min to see a snail brain to learn fighting 😂
pretty sure snails have WAY more neurons XD
"capable of balancing with ease"
*Agent continues to gyrate every limb uncontrollably*
But you gotta admit, that last dance-off was SICK 😂
"A ragdoll without a brain is merely a corpse; so lets give it a brain" Absolute BAR
Born to drunken fist,forced to box
10:10 this is literally how school fights are 💀💀
Sound track was perfect for the silliness 😂
It was amazing!!
5:27 he did a spinning kick that was awesome
0:44 "..but to ragdoll without a brain is merely a corpse" - my new favourite quote
this looks so funny its like newborns are learning to fight and walk at the same time
End result looks like a Florida parking lot fight.
8:26 i know what its like to be the gold guy in that scenario. Ive fought silver guy outside a club and just stood there like 'wtf are you doing bro?' While the guy took himself out 😂😂😂
wait.. I totally did not realise that you were such a small channel....the video quality is awesome!
I thought like he has 200k
You might not like it, but this is what peak combat performance looks like.
Cool music makes those random twitchings soo much epicier
Mf looks like he tryna rap battle with yellow at the end of the balance training
Dancing WHILE Boxing? Sounds like a mix of Boxing and Capoiera
Sounds like it but the end result is more spazoeira
You'd think so, wouldn't you, but actually its 2 puppets having seizures
I love how at the end of the video it becomes a techno dance battle/moshpit of two/drug addicts having a seizure
Great work! You put great effort in making the ai work as well as the editing of the video, keep it up!
So capeoira?
10:20 неиросеть настолько точно смоделировала бой двоих пьяных в стельку бомжей около моего подъезда .
How I imagine Jake Paul and Mike Tyson fight be like💀
Looking at the ragdoll model, it has a lot of "joints" for its hands but only two for the feet, our feet are basically hands as well, so it might help with learning to balance if you made the feet more complex as well.
Also, we are able to use our muscles with unequal emphasis stabilizing 0-100% on one leg alone, maybe there could be a mass X for the upper body which needs to be stabilized but can be shared unequaly between the legs, with shifting percentages.
Anyhow, they need way more standing and walking before they can fight. :)
7:20
When you and your buddy take the same amount at a festival but only one of you really feels it.
I hope this video gets the likes and views it deserves!
8:36 bro was getting aggressive
Excellent video and thank you for uploading in 4k60fps!
“Drunk people try fighting well progressively getting sober”
1:49 I’m sorry but u said observations complete wrong “ob-o-way-shins” is what u said 😭😭😂
I caught it the first time, but I'm just now realizing it.
I love seeing the starting frames of a superman punch get interrupted by a Q+E Sieger.
11:43 - Two drunk dads squaring up at a baseball game.
Loll
WtF … that’s not some new evolved combat technique. That’s two dudes having seizures.
General: What kinda combat robots were these?! They just kept dancing!
Robot: THREE TWO ONE ELECTRIC
Underrated. Commenting to boost algorithm
8:07 literally beating himself up
the sheer ammount of feints that these two are able to throw to each other, puts all profesional fighters to shame. Truly some high level stuff
They're becoming sentient!
What everyone else said. It is hilarious, and it is because there is no need to be economical with an energy reserve. Some cumulative refractory period on the muscles may help. It should learn to not exhaust them, so be static stable instead of dynamic stable.
I love the idea of a general AI looking back at this video and realizing we made it's ancestors fight each other for our amusement lol
Ai dancing and falling with absolutely no sounds exept for the background music is hilarious
"Today's mission..." Freakbait sounds different
This wasn't invented by a.i. Prince nazeem hamed was a master of this technique and many others had similar style.
I bet he gets to the cloud district very often
This gives an idea for a fight between 2 robots with ai operating them, with knowledge of all sorts of martial arts.. imagine one robot just uses a 540 kick to swiftly end the round.
A sentient being that you use as a slave, started to train for fighting but to your eyes is just a dance.
POV: you're a portuguese patron in 17th century Brazil
The AI right now just knows faster punching means bigger number which is why their movements were so much more wild before they could even stay up
Developer crates AI to optimize hand-to-hand combat, accidentally re-invents Capueira
Also 8:50 I'm pretty sure he's about to unleash a forbidden justu
I'd love to see a VR fighting game where it places you against a neural fighter. Sick documentary!
seems like they are reacting to each other too fast. human fight they react to each move but ai fight reacts to each line of code
they need some sort of energy system, no human would be able to move this fast without getting tired
Get one of these bots that do know how to fight and put it against the new bot that doesn’t know how to fight. Keep the simulation running until the new bot wins.
This is so high quality and you barely have 15k subs, youtube isn't doing justice.
They look like they're trying to fight while also trying to swat clouds ofmosquitoes
they might be getting in an imaginary fight like u can see one kicking in air and the other one is falling cuz of it
Was watching UFC the other day and briefly wondered what it would look like if you trained an AI to fight, and here it is! Guess Google was reading my mind again or something
This 3D stickman dancing kinda reminds me of Toribash.
The gravity is damped too much it should be set stronger also acceleration of musckes should risr linearly instead of "snap"
How is this music choice absolute perfection
I think a big part of the problem with the final product here is that during balance training, the bots learned to use their arms as integral parts of their balance; the punch training was awkward afterwards because they were still trying to use their hands to balance too much to really ever extend them meaningfully far from their own bodies. I don't necessarily know what the solution is - balance is just kind of a tricky thing to train- but some way to eliminate or reduce the allowed use of arms during balance training would probably allow them to throw punches more freely in later stages
Changing the training so that it rewards keeping their torso as close as possible to vertical could also help. And rather than random kick force to the hips, adding random forward momentum to the hands (and maybe torque to the hips, for twisting into the punch) could help them learn to compensate for punches, which could translate more effectively to the punch training stage
like what others have said. an "energy" cost per movement and force would be helpful, as it limits the movement necessary to keep balance, and also avoid erratic motions. I would also like to add that where a person is hit could affect their balance, especially the head, as that is where our sense of balance is being controlled. we can also observe and learn from our opponent's movement as well, and adapt accordingly. you could train the AI to learn from different fighting styles, give them limited energy, a sense of balance and movements to control it, and have them react to getting hit, or react to movements that would get them hit, sort of like an eye tracking/ object proximity detection. these are highly advanced and complex to simulate. but considering these factors would make you understand that there's more to fighting than movement alone
The first training session just looks like a cool dance party
It's like drunk capoeira without the kicks and twirls, and sped up...
The agent needs a response latency limit both in the perception and the reaction phase. As others stated, it also needs an energy cost that scales with movement size and force. Small movements like balancing or barely moving the arm to feint should cost almost nothing, large movements should cost a little, and consecutive high-effort movements should cost the most and add up to a diminished energy recovery over time. This energy system would be the single most complex part to program correctly, but if done right, it should result in much more realistic movements once you get the training balance right between energy conservation and pacifism.
*EDIT:* Also, their reaction strengths need to be imperfect. Lots of times fighters can overcompensate for feints, which then allows their opponent to capitalize with an actual attack, or sometimes even anticipate actions that never come, which usually ends up looking like erratic behavior to the opponent. But these anticipatory moves can also be used to lure the opponent into attacking, which can then allow a fighter to counter-capitalize on an opening that their fake reaction created.
Looks like two kids who challenged each other to fight after school but are scared to actually fight
"I'll sting like a bee and dance like AI"
The art of fighting without fighting
You should start by making them play tag. Take force out of the equation. 1 point for touching their forearm, and 3 points for touching their head, and 5 points for the chin. They should (at the very least) learn how to manipulate the opponents guard and create opportunities to land clean jabs.
Dude,I just couldn't... lmao! The setup was so serious and planned out then to witness that mess..omg had me dyin! 😂☠️
Idea: train the AI in a simulation, then make a robot irl with the exact same properties as the ai in the simulation and use your graphics card to make the ai behave the same way irl
Maybe reward it for taking 1 step a second that way it learns to step and fight as importance number 1. Once it takes 20 steps in 20 seconds then it knows that footwork is important as hell. But if you reward it more than once each second it might start tap dancing for needless points. Then teach it if it KOs its enemy it gets x10,000 points so the AI also tries to finish the fight instead of bouncing around for extra points.
I like the yellow fighter's plucky attitude and the strong rearward lean.
Do you also simulate exhaustion effects? I am pretty sure the result would be different. Why it is beneficial now? I would guess because it comes at no cost and the opponent is unable to 'read' your moves resulting in less hits received.
a little FREAKBAiT reference on the start there
So that's what it looks like when DaftPunk gets into an argument...
Some mentioned energy would help, I agree. Others said they're reacting to each other too fast. I would add to that, are they given a reflex window and response time for their actions? This could play into the energy system where they are given a response time, then allowed to speed it up or slow it down at the cost or conservation of energy drain, respectively. This would motivate their training towards playing slower and smarter. Additionally, to motivate aggression, penalize them for failing to strike, or getting hit. I believe this, coupled with a little more balance training that also uses energy and rewards an upright posture, would turn these agents into boxing legends.
the first training session reminds me of a game where there are like these people that like stumble and try to fight each other
The balance ai becomes so paranoid, scared of the invisible threats
Two Parkinson's patients try to beat eachother up
So basically, if you let an AI teach itself how to fight it just becomes the drunken master GOAT Emanuel Augustus
Would be pretty cool to see it learn different fighting styles from boxers. Imagine being able to put them in vr games so you could "fight" prime Mike Tyson, Muhammad Ali, Mayweather, pacman etc
AI:"You humans have unlimited stamina, right? Wait, what's this _fatigue_ thing?"
Yellow getting dropped by nothing is killing me
If you'd told me that these A.I.s were acoustic, I'd actually believe you
10:55
How I imagine deaf people have a rap battle
I wish instead of ending instantly, there was some sort of losing animation, like the losing boxer ragdolls or something
i think there was something missing from this: the human body wont really be hurt much at all by the very soft punches that the AIs are throwing, so i think there needed to be a minimum cap of force for a punch to count as well. this way they would only get rewarded if they threw a harder punch, which would probably end up looking a lot more "correct"
it also seems like they are waiting each other out for one to fall before the other sometimes instead of trying to hit. i think something was missing from the balancing training to make them more stable but im not entirely sure what, though maybe it was multiple different sources of balance to train off of, like each hour of training or something would pick a new random balance source.
surprisingly only the one source for a punch seemed to work fine, though, it does look like they adapted that into different kinds of punching like a hook and a jab and an uppercut.
is it possible to train a bot, to chase somthing an atack it? jump maybe? Then add a player controller after its fully trained and see how it will fight you?
Player controller sounds really nice. But the moveset will be too predictable and the AI should win easily
VR with motion camera may work
It appears that the AI is having a hard time trying to balance itself while mantaining the distance to the opponent. They throw occasional punches only when these 2 variables are OK and the punch won't make them fall...
0:00 "today's mission" FREAKBAIT