config_lqg.yaml

# Global seed for the RNG (-1 => random seed will be generated)
seed: -1

# Number of cores used for parallel path construction and evaluation
num_cores: 8

# Verbose output
verbose: False

# The path to the robot model file
robot_file: model/test_3dof.urdf

# The path to the environment file
environment_file: environment/env_3dof.xml

# The number of goal states generated by IK
num_generated_goal_states: 1

# Number of paths which have to be generated by RRT to evaluate
num_generated_paths: 500

# Timeout to generate 1 path (in seconds)
path_timeout: 100.0

# For the simplest toy problem. Use a linear path between the start and goal state
use_linear_path: False

# Number of simulation runs for the best found path
num_simulation_runs: 500

# Maximum angular velocity of the joints for the kinematic problem (in rad/sec)
max_velocity: 4.0

# The joint control rate
control_rate: 30.0

# The initial joint angles (has to be consistent with the number of links)
start_state: [0.7, 0.0, 1.5, 0.0, 0.0, 0.0]
#start_state: [0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0]
#start_state: [0.0, -1.57079632679489661923, 0.0, 0.0]

#Number of bins for the histograms
num_bins: 201


#################################################
## Uncertainty parameters
#################################################
# The error variables are defined as the standard deviation in percentage of the value range
# The minimum joint state covariance
min_process_covariance: 0.0

# The maximum joint state covariance
max_process_covariance: 5.0

# The minimum observation covariance
min_observation_covariance: 2.55

# The maximum observation covariance
max_observation_covariance: 5.0

#Determines which parameter (process covariance or observation
#covariance should be increased during simulation)
inc_covariance: process

# The number of covariance steps
covariance_steps: 5

path_deviation_cost: 200.0

control_deviation_cost: 1.0

################################################
## Reward parameters
################################################

# Reward params
discount_factor: 0.999999

# Penalty for crashing into an obstacle
illegal_move_penalty: 500.0

# Penalty for taking a step
step_penalty: 1.0

# Reward for reaching a terminal state
exit_reward: 1000.0

################################################

# The number of samples drawn from the distribution at each state to check a path for collision
sample_size: 1000

# Enforce the position and velocity constraints
enforce_constraints: True

# Enforce the control constraints
enforce_control_constraints: True

# Stop executing the path when state is terminal
stop_when_terminal: True

# Plot the paths generated by RRTConnect (should be 'False' for a large number of paths)
plot_paths: False

#Planning algorithm
planning_algorithm: RRTConnect

continuous_collision_check: True

#########################################
# Dynamic problem parameters
#########################################
dynamic_problem: True
gravity: 9.81
acceleration_limit: 100.0
simulation_step_size: 0.001

# Control sampler to use. 'discrete' or 'continuous'
control_sampler: discrete

# Number of controls to sample at the nearest node in the RRT tree
num_control_samples: 1

min_control_duration: 1
max_control_duration: 4

# The RRT goal bias
rrt_goal_bias: 0.05

#Determines if intermediate states along motions have to be added
#to the RRT tree (DON'T SET TO FALSE YET)
add_intermediate_states: True

#########################################
# Visualization
#########################################
show_viewer_evaluation: False
show_viewer_simulation: False

knows_collision: True