Reinforcement Learning (0.3.0)

Changelog

• UI/UX Improvements
  • Show additional danger-sense near boppie
  • Disabled spike rotation and blood in performance mode
  • Improved graphs by drawing lines instead of pixels in texture, graphs can be hidden
  • Neurons have colors based on their type, can now drag neurons
  • Show importance of neural network connections
  • Can now select different profiles for displaying neural network (weights or activations)
• Added lakes, water requirement for boppies and neurons for detecting lakes
• Boppie colors now represent: hue: dna, saturation: energy, luminance: water
• Added reinforcement learning, where neural net weights are adjusted based on rewards
• New neurons/senses
  • Terrain resistance sense below and ahead of boppie
  • Timer (time encoded in DNA)
  • Hunger/thirst
  • A single large water ray for detecting water
  • Ally sense for detecting allies (same as danger sense)
• Fixed major bug where each time a new boppie was spawned the AIs and NeuronTimers would become orphaned
• Required boppie offspring energy depends on DNA