Genetic Algorithm
Loading...
22,199
Loading...
Uploader Comments (draikolz)
Top Comments
-
cool robot. awful song.
-
love genetic...
see all
All Comments (15)
-
So genetic algorith is actually a AI that can learn by only using a chromosome with no programming?
-
Nice song!
Can you just copy and paste the song, I will search for it via Youtube
-
ah, so you're simply using an ANN and the GA is just finding the right weights. What are the inputs and outputs of the ANN? I love AI btw and I'm going to build a similar robot so I'm just gathering some ideas :)
-
love genetic~ ooooO~yeah
Loading...
Cool man, could you tell me how the chromosomes are encoded for this GA?
mtanti87 3 years ago
Its encoded by converting the weights, of each node into binary of 1 and 0 s. Then let GA manipulate the stings of bits and you'll get the new chromosomes that forms an individual.
eg you have 2 weights, each with a max of 8,
weight 1 is 4 weight 2 is 5.
Convert both of them and you get 100 and 101 so add them and you get 100101 then from there you have one chomosome. Let GA handle the rest of the crossover and mutations.
Good Luck :)
draikolz 3 years ago
intrusting video..
nice song, what is artist of this song?
Xeranas 3 years ago
i dont know sorry.... i cant read the characters... huhu
draikolz 3 years ago
Cool! but in the end it looks just as smart as it started, except for that one time when it was rocking back and forth. How'd you evaluate the fitness function for all the individuals with only one robot? Did you write a simulation?
domokato 4 years ago
No theres no simulations done with this try. It was more of a try and error. The fitness is a basic one for avoiding walls... Best try which has the most distance and least hitting the wall has best fitness. Each try(individuals) are as long as 6 seconds... after 10 tries... enough population experience is attained and from there the next population is evolved.
draikolz 4 years ago