racesim_config.yaml

#! configure environment
QLEARNING:
  REWARD_FUNCTION: # TODO: somehow store a function to use
    - none
  ENV_EPISODES: 1
  ENV_MAXTRY: 1000
  ENV_EPSILON: 1
  ENV_EPSILONDECAY: 0.999
  ENV_LEARNINGRATE: 0.1
  ENV_GAMMA: 0.6
  ENV_OBSERVATION_FIELDS:
    - race_position
    - car_fuelMass
  ENV_OBSERVATION_LOW:
    - 1
    - 0
  ENV_OBSERVATION_HIGH:
    - 15
    - 100

#! configure race simulation
RACESIMULATION:
  RACE_LENGTH: 27
  RACE_GRIDSIZE: 15
  RACE_INITFUELMASS: 100.0

#! configure models
MODELS:
  PITSTOP:
    TIMELOSS_TRAVELIN: 30.0
    TIMELOSS_TRAVELOUT: 45.0
    STANDINGTIME_FREE: 25.0
    STANDINGTIME_REGULATION: # TODO: integrate mandatory times from regulation
      - none

  OVERTAKING:
    OVERTAKE_SUC_CAR1: 0.5
    OVERTAKE_SUC_CAR2: 1.5
    OVERTAKE_UNS_CAR1: 3
    OVERTAKE_UNS_CAR2: 1

  RACESTART:
    TIMELOSS_BASE: 5.0

#! configure logging
LOGGING:
  ENVIRONMENT: false # logs for all environment events (starting episode, finished episode)
  EPISODE_INTERVAL: 1 # after how many episodes the agent results should be shown (recommended 50)

  SIMULATION:
    DONE: false # logs that the done() function returned true
    LAP: false # this logs the currently simulated lap count before each lap simulation
    SECTOR: false # this logs the currently simulated sector count before each sector simulation
    GRID_POSITIONS: true # logging a table of lap+sector timings after each lap
    GRID_GENERATION: false # logging a table of the grid after generation
    POSITION_CHANGES: false
    OVERTAKES: false # this will print every overtake

  AGENT:
    ACTIONS: false # log the action taken by the agent and the calculated reward after each lap
    RESULTS: false # this displays the agents results (race_position, total_reward) after each episode

  DUMP: false # this will dump verbose logs of all participants after each episode