Search For Powder Intelligence returns! by perogiepro

perogiepro

28 / 7

22nd Dec 2015

10th Jan 2016

We are back, refurbished and with new ideas. Read all of our old discussion on the forum under a topic called Evolution in a video game! Anyways the main plan is to create small bots that eat to survive, reproduce, and then change (genes). Check list.

sfpi intelligence evolution electronics test group science life workinprogress

Comments

• 

  Windspren 31st Dec 2015
  perogiepro: Did you see my idea of a bot that has the "brain" outisde of it? The bot actually works.
• 

  Windspren 31st Dec 2015
  Gel genomes sound like they would be better, as we can have more combinations of genes. Do you have an example bot?
• 

  msasterisk 31st Dec 2015
  Perogiepro: Interesting idea, sexual reproduction in PSTN robots. It could work! I think we may not use gas genomes, but look at my previous comments. I had some ideas...
• 

  perogiepro 31st Dec 2015
  Here's the design idea. Probably needs alot of tinkering, like how to make the bot still able to move. ID:1914438
• 

  perogiepro 31st Dec 2015
  Amazing! And for replicating I think we just have one gender bot have a side with DRAY and a way to beam (dray again) it's gases (to pass on genes) over to the new bot. So what happens is the DRAY gender lines up with a bot that can activate it, then the dray bot replicates the bot to it's right and both cells dray their gases over. Ill make a save that visualy demonstrates it.
• 

  msasterisk 31st Dec 2015
  Also, MWAX randomizers could be useful. They could be toggled between steadily sparking one output and randomly sparking outputs, to control whether the bot goes in a straight line or just derps around.
• 

  msasterisk 31st Dec 2015
  Oohhh, nice idea! If we can get Schmolen's awesome neural net inside a bot, awesomeness would happen!
• 

  Karakanlud 31st Dec 2015
  I'm experimenting with a pseudobrain right now. Instead of randomisers, it has heat mediated cells (like the ones in Schmolen's saves). It would operate with an adjustable AI relaxed signal (maybe it would be connected some way to sensors to make it "change its mind". Also planning to add a sensory imput, with a small short-term memory unit, which would make it remember the things detected in the last 10-15 seconds.
• 

  12Me21 31st Dec 2015
  Is anyone going to make a SFPI group? (and maybe a group account, which is legal as long as it isn't used to vote on anything).
• 

  msasterisk 30th Dec 2015
  I've got it! GEL! A lower tmp makes it stick to the sides of the tube it's in, while a high tmp makes it act more and more like WATR. At tmp 100, it is basically a MERC randomizer, but with no SPRK freezing. Running it through Sandwichlizard's lab, I found that the randomness is pretty much proportional to tmp.