Try the following link: https://nrsharip.github.io/ai-genetic-algorithm-snake/
This is a port of AI-for-Snake-Game (see The AI for Snake Game Chronicles by Craig Haber) from Python to Javascript.
UPDATE: Managed to reach the score of 79 by adjusting the criterias (see RUN 2022-07-24 21:21:26).
NOTE: To accelerate the training process use the multiprocessing python version.
Pareto Fronts and Scatter Charts
Generation #4140:
 |
 |
Criterias changed:
(see this line in snakeGameGATest.js)
if self.frames_since_last_fruit == 100:
self.game_over()
(see this line in snakeGameGATrain.js)
if self.frames_since_last_fruit == 100:
frame_score = self.frames_alive - self.frames_since_last_fruit
if frame_score <= 0:
frame_score = 0
(see this line in snakeGameGATrain.js)
_1 = self.score**(4.5 - (self.score + 1)**(1/5))
_2 = frame_score**((self.score + 1)**(1/3))
fitness = _1 * _2
RESULTS:
(see GAdata.txt):
Generation 4007 Best Game Score: 79
Generation 3186 Best Game Score: 73
Generation 3465 Best Game Score: 73
Generation 4141 Best Game Score: 73
Generation 3448 Best Game Score: 72
Generation 4035 Best Game Score: 72
Generation 4223 Best Game Score: 72
Generation 4368 Best Game Score: 72
Generation 4449 Best Game Score: 72
Generation 3825 Best Game Score: 71
Generation 4077 Best Game Score: 71
Generation 4332 Best Game Score: 71
Generation 4333 Best Game Score: 71
 |
 |
Generation #4140 (4152) CPG:200 BPW:8 I:9 H:10 O:4 failures:
 |
 |