Learning to use firearms aginst somthing that shoots back.
@Zuzelo Жыл бұрын
@@Etvald Hmmm, I have a similar project where I trained AI soldiers to fight eachother and I was planning on coming back to that project. So I think I might do that :)
@MyDude199 Жыл бұрын
Chiklz flying a plane
@Aunixi Жыл бұрын
Ai learns how to play catch
@TheOriginalRX Жыл бұрын
@@Zuzelo Chiklz tries to get a girlfriends. Ya boy is lonely
@alansmithee419 Жыл бұрын
5:33 Reminds me of when we did a spear-throwing thing on a school trip once. Most people either missed or hit this foam mock-up of a boar in the foot. I somehow managed to lodge it in its neck, which came as a surprise to everyone, most of all me. Was quite funny. Thank you for coming to my ted talk.
@Zuzelo Жыл бұрын
Thank you for giving this inspiring ted talk :D
@Aunixi Жыл бұрын
I just love how pogo wiggles at 50m
@Jacobconnor52511 ай бұрын
37 is my favorite number
@Spencyshorts9 ай бұрын
Underated!!
@Natural_Power Жыл бұрын
Why is there a raycast and not just a known distance though? I feel like the raycast doesn't do much since only distance is variable
@Zuzelo Жыл бұрын
Besides the fact that it is cooler this way, it is because initially I planned on having vertical rotation as well, so chiklz would need to know where exactly the target is. But since I removed it, it would definitely make more sense to use the distance instead.
@amegatron07 Жыл бұрын
What a time to be alive! I love how the topic of RL is becoming more popular, and how such frameworks as ML-Agents allow people to come up with numerous funny ideas)))) And most often it is really so funny to watch those agents trying to learn something) P.S. Btw, giving higher reward for a farer target is not really relevant here in the context of training, because that's not what the agent can choose or progress to by itself.
@Zuzelo Жыл бұрын
Agreed and glad you enjoyed the video :D As for the reward, it wasnt based on target distance, it was higher if the arrow fell close to the target. Closer the arrow falls to the target - bigger the reward.
@amegatron07 Жыл бұрын
@@Zuzelo thanks for clarification) My bad)
@EgalIlI Жыл бұрын
I love your videos, thank you for these wonderful ai simulations
@Zuzelo Жыл бұрын
Thanks for watching :)
@kinosaurs5155 күн бұрын
5:33 XDD
@Emperors_A7MET Жыл бұрын
I use to think about it. these are all the ideas i think about but don't have skills enough to implement it. Lol
@Zuzelo Жыл бұрын
Haha, I would love to hear those ideas either here or discord! Who knows, perhaps next video will be about one of them ;)
@Emperors_A7MET Жыл бұрын
Ok i am on your discord
@krisztiankoblos1948Ай бұрын
So you fit a parabolic function with 2 parameters (sounds so creazy ) :P ?
@kinosaurs5155 күн бұрын
quiz answer: because his raycasts are 50m long I guess?
@balltickler-hm5mo12 күн бұрын
I understand that pogo is a funny bean dude but what is chiklz?
@krisztiankoblos1948Ай бұрын
Do the AI knows the distance or it is massured by visuals?
@MyDude199 Жыл бұрын
Man i thought C was the right answer.
@Zuzelo Жыл бұрын
Haha, you overestimated me xD
@lamp774611 ай бұрын
I wish there was a little more humor in the videos.
@nikolozgilles Жыл бұрын
robin hood bot :)
@Zuzelo Жыл бұрын
But without the stealing part :D
@nikolozgilles Жыл бұрын
actually, william tell! :D
@iuliusconnor6918 Жыл бұрын
You see, if only you made him human and not lizard-thing, he would have had real eyes and therefore seen beyond 50m :) classic mistake
@Zuzelo Жыл бұрын
Y u bully Chiklz?
@iuliusconnor6918 Жыл бұрын
@@Zuzelo Chiklz silly
@lemonadecat482211 ай бұрын
Small d = big reward Big d = small reward
@VascoCC9511 ай бұрын
Why have you used ray-casting in this case? It made no sense... It would be better to use the distance or if you wanted to give Chiklz a challenge you could give him the perceived size of the target to give a variable increment value as the distance and make him tackle that (as that's the case with us humans, we can only estimate the distance based on the size that we see vs the size we expect...)