Interesting Idea

[Tj3]

A bot that relies almost entirely on evolution and mutations, and starts with only core genes... But with like maybe multiple of the cores, to preserve important functions in the beginning? With .mrepro in all or some conditions, to force mutations, especially when doing bad... Maybe a chance of mrepro? Maybe a bot that replicates very often, but using .mrepro unless performing excellent, so once you get a good bot evolved, it stops mrepro'ing and quickly duplication... This could be very interesting now than I think of it...
I wonder how complex they could evolve...

[theblaze]

good idea

[Panda]

I could say that it would be quite easy to implement, assuming that if it is doing well, it would have a high population.

[Botsareus]

I already started that one:


http://www.darwinbots.com/Forum/index.php/topic,3519.msg1383572.html#new


I am still working it, expect more results soon.

[Botsareus]

I know this is a really old post but I decided to kick it up a notch, after we release the chloroplast version, I will be developing a version of DB to.... drums...  evolve a zero bot into a robot that can survive fairly well in F1 mode. 

[rwill128]

Hey,

I just discovered DB a couple weeks ago, but it's nice to see that people are still using these forums.

I've been curious what kind of results people are getting with new machinery, especially since some of these posts are over 5 years old.

I'm especially curious about how large of a sim people can run on new hardware, and how quickly it runs, etc.

---

Also, his idea sounds really interesting. What have people come up with when trying to create a stable base for a bot that is intended to acquire skills through mutation? I like his idea of having several versions of the gene that has core functions, and having each of those genes either run completely if it's doing well, or .mrepro if it's not.

Has anyone built more sophisticated status checks into the genes, also, maybe a beginning and ending gene that not only carries out core functions, but also uses some kind of check to ensure that the core functions have been preserved, and resets them if they haven't been. There could be a middle gene full of code that's essentially blank (functionless), but t takes up space so that when mutations occur, they're likely to occur in this block of code.

I'm new to the language, so I don't know if what I just suggested is possible or not.

[Botsareus]

I've been curious what kind of results people are getting with new machinery

Unfortunately, the performance has not improved much since vb6 (the language darwinbots is written in) can not do multi-threading

What have people come up with when trying to create a stable base for a bot that is intended to acquire skills through mutation?

Go to http://wiki.darwinbots.com/w/Botsareus%27s_Vision

Unfortunately, this has been written for an older version of darwinbots (2.44.1). I will try to merge this into a newer version, but as you can see I got stuck with finishing up other ideas such as the crossover fix and chloroplasts.