diff --git a/examples/hpo/quadrotor/config_overrides/ilqr_quadrotor_2D_attitude_hpo.yaml b/examples/hpo/quadrotor/config_overrides/ilqr_quadrotor_2D_attitude_hpo.yaml index 8a8e8e502..e4c457a61 100644 --- a/examples/hpo/quadrotor/config_overrides/ilqr_quadrotor_2D_attitude_hpo.yaml +++ b/examples/hpo/quadrotor/config_overrides/ilqr_quadrotor_2D_attitude_hpo.yaml @@ -1,9 +1,9 @@ hpo_config: load_if_exists: True # this should set to True if hpo is run in parallel - objective: [exponentiated_rmse] # [other metrics defined in base_experiment.py] - objective_bounds: [[0.0, 1.0]] # [bounds for each objective]. Worse value will be assigned if objective evaluation is None - direction: [maximize] # [minimize, maximize] + objective: [exponentiated_rmse, exponentiated_rms_action_change] # [other metrics defined in base_experiment.py] + objective_bounds: [[0.0, 1.0], [0.0, 1.0]] # [bounds for each objective]. Worse value will be assigned if objective evaluation is None + direction: [maximize, maximize] # [minimize, maximize] repetitions: 5 # number of samples of performance for each objective query n_episodes: 5 # number of episodes to evaluate each policy use_gpu: True diff --git a/examples/hpo/quadrotor/config_overrides/linear_mpc_quadrotor_2D_attitude_hpo.yaml b/examples/hpo/quadrotor/config_overrides/linear_mpc_quadrotor_2D_attitude_hpo.yaml index 4f9ace335..668a302f4 100644 --- a/examples/hpo/quadrotor/config_overrides/linear_mpc_quadrotor_2D_attitude_hpo.yaml +++ b/examples/hpo/quadrotor/config_overrides/linear_mpc_quadrotor_2D_attitude_hpo.yaml @@ -1,9 +1,9 @@ hpo_config: load_if_exists: True # this should set to True if hpo is run in parallel - objective: [exponentiated_rmse] # [other metrics defined in base_experiment.py] - objective_bounds: [[0.0, 1.0]] # [bounds for each objective]. Worse value will be assigned if objective evaluation is None - direction: [maximize] # [minimize, maximize] + objective: [exponentiated_rmse, exponentiated_rms_action_change] # [other metrics defined in base_experiment.py] + objective_bounds: [[0.0, 1.0], [0.0, 1.0]] # [bounds for each objective]. Worse value will be assigned if objective evaluation is None + direction: [maximize, maximize] # [minimize, maximize] repetitions: 5 # number of samples of performance for each objective query n_episodes: 5 # number of episodes to evaluate each policy use_gpu: True @@ -14,6 +14,18 @@ hpo_config: # hyperparameters hps_config: - horizon: 40 - q_mpc: [5.0, 0.1, 5.0, 0.1, 0.5, 0.001] - r_mpc: [0.8, 0.8] + horizon: 25 + q_mpc: [15, 0.1, 15, 0.1, 0.5, 0.01] + r_mpc: [15., 5.] + vizier_hps: + horizon: 30 + q_mpc: + - 14.315870549815472 + - 2.5426733145296656 + - 15.0 + - 0.0001 + - 0.0001 + - 0.0001 + r_mpc: + - 15.0 + - 6.471762097184253 diff --git a/examples/hpo/quadrotor/config_overrides/lqr_quadrotor_2D_attitude_hpo.yaml b/examples/hpo/quadrotor/config_overrides/lqr_quadrotor_2D_attitude_hpo.yaml index c83b92b0b..b6e9cf33f 100644 --- a/examples/hpo/quadrotor/config_overrides/lqr_quadrotor_2D_attitude_hpo.yaml +++ b/examples/hpo/quadrotor/config_overrides/lqr_quadrotor_2D_attitude_hpo.yaml @@ -18,15 +18,15 @@ hpo_config: r_lqr: [1.0, 1.0] vizier_hps: q_lqr: - - 14.974836994854666 - - 1.6803329734409302 - - 0.02535896270931672 - - 0.00028863807214684776 - - 1.1108878724895586 + - 5.88529216699797 - 0.0001 - r_lqr: - - 0.0003546601186005271 + - 5.00397782633266 + - 0.8618608985543565 + - 0.980013664114415 - 0.0001 + r_lqr: + - 1.2497823191792372 + - 0.8638794170664649 optuna_hps: q_lqr: - 5.0 diff --git a/examples/hpo/quadrotor/config_overrides/pid_quadrotor_2D_attitude_hpo.yaml b/examples/hpo/quadrotor/config_overrides/pid_quadrotor_2D_attitude_hpo.yaml index e87117a44..cb363fab5 100644 --- a/examples/hpo/quadrotor/config_overrides/pid_quadrotor_2D_attitude_hpo.yaml +++ b/examples/hpo/quadrotor/config_overrides/pid_quadrotor_2D_attitude_hpo.yaml @@ -23,51 +23,34 @@ hpo_config: d_coeff_tor: [20000., 20000., 12000.] vizier_hps: d_coeff_for: - - 2.129021744940595 - - 0.4640573281692209 - - 2.542887187784077 + - 0.1890092870696659 + - 1.0541122664300269 + - 1.1247264394469974 d_coeff_tor: - - 14998.998579124276 - - 10019.881571806687 - - 10534.570451701942 + - 14624.865780764232 + - 14735.949237321409 + - 12928.8956036557 i_coeff_for: - - 0.0014889448824393103 - - 2.1744158095428565 - - 0.0029573499100777104 + - 0.001 + - 0.4045145572372126 + - 0.6961573899435679 i_coeff_tor: - - 441.54939925530545 - - 0.5010238470068967 - - 444.3856421179262 + - 0.0 + - 7.906897732216709 + - 524.6539429699372 p_coeff_for: - - 4.886838283343341 - - 4.608092291879853 - - 4.996853766830539 + - 0.4458780003496436 + - 2.118873126448762 + - 1.2386787893322404 p_coeff_tor: - - 50002.195370586196 - - 79982.3047902415 - - 50006.91012411862 + - 64299.26234435173 + - 73880.50880762804 + - 59700.826209375475 optuna_hps: - d_coeff_for: - - 1.5604761716915811 - - 1.3593133298420035 - - 3.2553437457217855 - d_coeff_tor: - - 12782.496926017255 - - 13315.217714132743 - - 13650.965688467855 - i_coeff_for: - - 2.4249337311042356 - - 0.5072768704878523 - - 4.280513008919815 - i_coeff_tor: - - 518.9194229378894 - - 298.29551117414286 - - 332.42926926363987 - p_coeff_for: - - 4.553904304026884 - - 3.439182803233737 - - 4.778645189386438 - p_coeff_tor: - - 68371.22038933045 - - 69727.19131379195 - - 68354.19711764912 + p_coeff_for: [.4, .4, 1.25] + i_coeff_for: [.05, .05, .05] + d_coeff_for: [.2, .2, .5] + + p_coeff_tor: [70000., 70000., 60000.] + i_coeff_tor: [.0, .0, 500.] + d_coeff_tor: [20000., 20000., 12000.]