I regularly practice Aikido since 1989: when I have the chance I always like to practice or check out videos from HEMA(1) because even if most of the weapons are taught differently, I find fascinating both the very common ideas around dagger or unarmed combat, and anyway understanding what the differences are and why these exist is always very interesting.
I'm a Fiore scholar-level longsword practitioner and cannot say enough amazing things about learning to use weapons like this based on historic manuscripts. It's why my HN username is "waster" – we train from the start with wooden wasters (practice swords), as was done historically. If you're in the SF Bay Area and care to learn, check out swordfightingschool.com; or search for a HEMA (Historical European Martial Arts) school or club near you.
Hello fellow "waster"! It's nice to see another one from the other side of world! I practice the Bolognese fighting school (Achille Marozzo and Antonio Manciolino) at our fencing club (Sala d'Arme Achille Marozzo). Fun fact: I was born, raised, and currently living and writing from Premariacco, Fiore's birthplace :)
well ... it would be interesting, but i fear that an AI robot vs. human swordfighting style would probably be really boring (given that it'd be possible to make a robot with human attributes, i.e. strength, sensory input, etc.). my guess is a very small set of moves would emerge that are, for practical purposes, un-counterable, depending on the proportions of the particular human opponent.
that said, how to train the AI? robot to human test battles are very limited in training set size, so it would have to be AI robot vs. AI robot. that would probably yield a strategy very close to the style of the AlphaGo/zero/*: very conservative, maximizing the space of winning scenarios, winning each one with as little of an advantage as necessary.
edit: i've got not the slightest clue about sword fighting (except that it doesn't look like in the movies), so i could be completely off the mark here.
Given the convergent evolution of fighting styles around the world, I think the result would be pretty familiar. By far the tricky part would be evaluating a "style": in real life, footwork and using biomechanics to eke out a tiny edge in speed play a role. And of course you have to start over moving to a different type of sword, or a different opposing weapon...
10 comments
[ 8.0 ms ] story [ 47.8 ms ] threadOther historic fencing manuals like this can be found at https://wiktenauer.com
https://youtube.com/playlist?list=PL2AcjEcWawSm_rc8F_bLu50aP...
1) https://en.wikipedia.org/wiki/Historical_European_martial_ar...
that said, how to train the AI? robot to human test battles are very limited in training set size, so it would have to be AI robot vs. AI robot. that would probably yield a strategy very close to the style of the AlphaGo/zero/*: very conservative, maximizing the space of winning scenarios, winning each one with as little of an advantage as necessary.
edit: i've got not the slightest clue about sword fighting (except that it doesn't look like in the movies), so i could be completely off the mark here.