TPT evolution? by perogiepro

perogiepro

99 / 13

24th Aug 2015

22nd Dec 2015

We are the SFPI, we are searching for mechanical life that can detect danger and think. essentialy basic AI. Our end goal is to make one that can replicate and change, so it will """evolve"""

machine intelligence experiment neuron life survival challange evolution spfi

Comments

• 

  RCAProduction 26th Aug 2015
  FightinJack seems to have done it.
• 

  RCAProduction 26th Aug 2015
  Just like before, the bot just breaks itself when it goes two directions at once. That needs to be fixed.
• 

  FightinJack 26th Aug 2015
  I fixed your brain area so that it doesnt make the pstn box bot malfunction... you can use it here: id:1845034
• 

  Technomancer 26th Aug 2015
  The problem with multiple robots is that, even if you consider them as cells of a larger organism, the 'brain' doesn't have the ability to coordinate them. Several 'cells' might all detect the same object, but depending on their relative positions they may perceive it to be in different directions. I think this is what happened when I tried copy-pasting the robot next to itself. The two cells made about six moves together, then tore themselves to pieces in perfect synchronization.
• 

  Technomancer 26th Aug 2015
  @Classeh, you can control WIFI temperatures either by using CRAY rather than DRAY to spawn the WIFI, or by EXTREMELY precise use of protons (which alter the temp of anything they pass through, including electronics that can't normally gain or lose heat). The latter method would be horribly difficult to implement though.
• 

  Technomancer 26th Aug 2015
  You can probably get around the sensory overload issue by expanding the logic a little bit to effectively reduce the sense input to a simple 'did ANY unit detect xxx in a given direction? Did NO unit detect yyy in that same direction?'. Then if both answers are 'yes' for more than one direction, either choose a direction based on number of sense inputs received, or simply randomize between acceptable directions.
• 

  Schmolendevice 26th Aug 2015
  Well to note, my present neural network does in ways have the capacity for 'memory' or to be able to continue moving in the absence of stimulus. From my understanding of neural mechanisms, there is nothing random about it except the environment. But indeed, I see that the thing here is sort of working. My design at present never outputs more than one motor signal at a time.
• 

  classeh 26th Aug 2015
  We shoudl start using INST. It's faster
• 

  Schmolendevice 26th Aug 2015
  hmm, so essentially my bot is too slow
• 

  perogiepro 26th Aug 2015
  but it would have to be big. and with that i have to go to bed. good night everyone. keep at it we have come a long way. and feel free to post your own creations for the SFPI group to consider and use