config_data_gen.yml

controller: mpc
ML_Pipeline_mode: True  # If True a directory structure required by SI_Toolkit will be created, location taken from config_training, not from here.
split: [0.8, 0.1]  # Train/Val/(Test) split - only matters if you run it in ML Pipeline mode
PATH_TO_EXPERIMENT_RECORDINGS_DEFAULT: './Experiment_Recordings/'   # Where to save experiment recording per default, can be overwritten in config_training if ML_Pipeline_mode is True
seed:   # If not assign random seed based on datetime is used
length_of_experiment: 360  # Length of each experiment in s:
random_initial_state:
  # Unassigned variables will be randomly initialized (see below)
  position:
  positionD:
  angle:  #Seems to be in radian
  angleD:
  target_position:
  init_limits: # The initial state is randomly chosen from below range if above is not assigned
    angle: [0.0, 180.0]  # degree, 0 is up, 180 down, set the range for right half plane, same will be applied to left
    angleD: 1200.0
    position: 0.8  # Fraction of TrackHalfLength to each side
    positionD: 0.5 # Fraction of TrackHalfLength to each side
start_at_target: True  # initial target position should be equal to initial state
track_fraction_usable_for_target_position: 1.0  # Set the max for smoothly interpolated random target position to avoid bumping into track ends.
target_position_end: ~  # value OR not assigned - random value is taken
initial_target_equilibrium: 'up'  # possible 'up', 'down', 'random' = 50/50
keep_target_equilibrium_x_seconds_up: 10
keep_target_equilibrium_x_seconds_down: 2.5
dt:
  simulation: 0.002  # simulation timestep, s
  control: 0.02  # control rate, s
  saving: 0.02  # save datapoints in csv in this interval, s
turning_points:
  track_relative_complexity: 1  # Randomly placed target points/s
  interpolation_type: '0-derivative-smooth'  # How to interpolate between turning points of random trace, Possible options: '0-derivative-smooth', 'linear', 'previous'
  turning_points: # List of target positions, can be None to simulate with random targets, Example: turning_points_DataGen = [0.0, 0.1, -0.1, 0.0]
  turning_points_period: 'regular' # How turning points should be distributed, Possible options: 'regular', 'random'; never used, leave it as it is
save_mode: 'online'  # It was intended to save memory usage, but it doesn't seems to help, setit to "offline" only if you want to show summary plots
# Show popup window in the end with summary of experiment?
show_summary_plots: False
show_controller_report: False
number_of_experiments: 12  # How many experiments will be generated