config.yaml

mode:
  algo: CANONICAL_ES #Valid Options: 'DDPG', 'ES', 'DQN','ADAM_ES' 'HARDCODED'
domain:
  folder: ../Domains/
  warehouse: Small_SS #Valid Opt: 'Small_SS', 'Tiny_SS', 'Min_SS', 'Medium_SS'
  agents: link #Valid Opt: 'centralized', 'centralized_t','centralized_avgt', 'centralized_botht', 'link', 'link_t', 'link_avgt', 'link_botht', 'intersection', 'intersection_t', 'intersection_avgt', 'intersection_botht'
DDPG: #config options for 'DDPG' leanring algorihm
  epochs: 10
  runs: 1 # number of statistical runs
  rand_proc_std_dev: 0.2 #standard deviation of the random process, higher values encourage more exploration
  batch_size: 50 
ES:
  epochs: 500
  runs: 30 # number of statistical runs
  population_size: 1000 #1000 is the default size
  learning_rate: 0.001 
  rand_proc_std_dev: 0.2 #standard deviation of the random process, higher values encourage more exploration
CANONICAL_ES:
  epochs: 500
  runs: 30 # number of statistical runs
  population_size: 1000 #1000 is the default size
  learning_rate: 0.001 
  rand_proc_std_dev: 0.2 #standard deviation of the random process, higher values encourage more exploration
ADAM_ES:
  epochs: 500
  runs: 30 # number of statistical runs
  population_size: 1000 #1000 is the default size
  learning_rate: 0.01
  rand_proc_std_dev: 0.5 #standard deviation of the random process, higher values encourage more exploration
  b1: 0.9 # Adam_ES momentum exponential weight defined in [0,1]
  b2: 0.9999 # Adam_ES dampening exponential weight defined in [0,1]
DQN:
  epochs: 200
  runs: 1 # number of statistical runs
  rand_proc_std_dev: 0.2 #standard deviation of the random process, higher values encourage more exploration
  # batch_size: 3   
simulation:
  verbose: true
  steps: 200
  agvs: origins_200.csv #Valid options: 70,90,120,200,400
  goals: goals.csv
results:
  folder:     ../Results/
graph: # do not change this unless you know what you are doing
  vertices:   vertices.csv
  edges:      edges.csv
  capacities: capacities.csv