parameter_scheduling_mpc

Finds the optimal parameter scheduing policy of MPC for differential-drive wheeled mobile robot by solving MDP via dynamic programming

Dependencies

• CasADi: https://web.casadi.org/

Usage

python3 solve_mdp.py

Optional arguments

• env-dt: type=float, default=0.05
• env-pose_tol: type=float, default=1e-2
• env-control_tol: type=float, default=1e-2
• env-timeout: type=float, default=1000
• mpc-dt: type=float, default=0.05
• mpc-time_horizon: type=int, default=10
• mpc-regularization_control: type=float, default=1e-3
• mpc-regularization_v: type=float, default=0
• mpc-regularization_omega: type=float, default=0
• mpc-r_tol: type=float, default=5e-3
• n_r: type=int, default=40
• n_alpha: type=int, default=24
• n_psi: type=int, default=25
• n_theta: type=int, default=7
• repeats_per_sample_point: type=int, default=1
• mdp-discount: type=float, default=0.99
• mdp-max_iter: type=int, default=100000
• mdp-tol: type=float, default=1e-6
• mdp-solver: type=str, default='ValueIteration', choices=['PolicyIteration', 'ValueIteration']
• data_dir: type=str, default='data/mdp'
• sample: type=bool, default=True
• solve: type=bool, default=True
• test: type=bool, default=True