Genetic Algorithm Walkers

"It's like watching other people QWOP"


Current Generation (0)

Name Score

Record History

Gen Name Score


Gene mutation probability
Gene mutation amount
Champions to copy
Motor noise
Round length
Animation quality
Simulation speed

Lower the animation quality and simulation speed to save some CPU.

Zero frames per second and high simulation speed will give you the fastest results, but processor usage will go up as well.


What the hell is this?

This observational pastime hopes to evolve walking creatures through genetic algorithms.

Who comes up with these names?

The names are generated based on each creature's genome. Since the genetic algorithm tends to produce creatures with similar genes, two creatures with similar names will have similar traits.

Sometimes two creatures can have the same name by coincidence, as there are nearly infinite genome possibilities and limited numbers of letters in each name. Much like two people in real life with the same name, that doesn't mean they will behave the same.

Apologies in advance for any dirty words spelled by accident. Bad genome. :)

How is the score calculated?

In order to promote upright walking, each creature receives points based on how high the head is in relation to its feet, and how much it advances while upright. Bonus points are given for each proper step forward.

Can I give you riches and wealth?

Yes, but I'll use them to buy coffee.

Who made this?

This exercise in pointless science is brought to you by (which is really just one guy).

Physics engine brought to you by the amazing Erin Catto, I'm using the very nice JS port by Paril.

Additional code by Leif Johnson.

It looks broken on my browser!

Your browser broke my toy! Make it pay!

I'm a Computer Science snob and your approach to genetic algorithms is flawed.

But it's fun, ain't it? :)

I hate boxy humanoids, can I have misshapen cars instead?

Glad you asked!