Reinforcement learning class#

The following are examples of reinforcement learning methods for robot learning.

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_Ant_RofuncRL.py --agent=[ppo|a2c|td3|sac]

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_Ant_RofuncRL.py --agent=ppo --inference

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_CURICabinet_RofuncRL.py --agent=ppo

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_CURICabinet_RofuncRL.py --agent=ppo --inference

# Available objects: Hammer, Spatula, Large_Clamp, Mug, Power_Drill, Knife, Scissors, Large_Marker, Phillips_Screw_Driver

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_DexterousHands_RofuncRL.py --task=CURIQbSoftHandSynergyGrasp --agent=ppo --objects=Hammer

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_DexterousHands_RofuncRL.py --task=CURIQbSoftHandSynergyGrasp --agent=ppo --inference --objects=Hammer

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_FrankaCabinet_RofuncRL.py --agent=ppo

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_FrankaCabinet_RofuncRL.py --agent=ppo --inference

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_FrankaCubeStack_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/IsaacGym_RofuncRL/example_Humanoid_RofuncRL.py --agent=ppo

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_Humanoid_RofuncRL.py --agent=ppo --inference

'''Training'''
# Backflip
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_backflip --agent=amp
# Walk
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_walk --agent=amp
# Run
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_run --agent=amp
# Dance
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_dance --agent=amp
# Hop
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_hop --agent=amp

'''Inference with pre-trained model in model zoo'''
# Backflip
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_backflip --agent=amp --inference
# Walk
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_walk --agent=amp --inference
# Run
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_run --agent=amp --inference
# Dance
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_dance --agent=amp --inference
# Hop
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidAMP_RofuncRL.py --task=HumanoidAMP_hop --agent=amp --inference

'''Training'''
# Getup
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEGetupSwordShield --agent=ase
# Getup with perturbation
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEPerturbSwordShield --agent=ase
# Heading
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEHeadingSwordShield --agent=ase
# Reach
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEReachSwordShield --agent=ase
# Location
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASELocationSwordShield --agent=ase
# Strike
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEStrikeSwordShield --agent=ase

'''Inference with pre-trained model in model zoo'''
# Getup
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEGetupSwordShield --agent=ase --inference
# Getup with perturbation
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEPerturbSwordShield --agent=ase --inference
# Heading
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEHeadingSwordShield --agent=ase --inference
# Reach
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEReachSwordShield --agent=ase --inference
# Location
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASELocationSwordShield --agent=ase --inference
# Strike
python examples/learning_rl/IsaacGym_RofuncRL/example_HumanoidASE_RofuncRL.py --task=HumanoidASEStrikeSwordShield --agent=ase --inference

'''Training'''
# Available tasks: BiShadowHandOver, BiShadowHandBlockStack, BiShadowHandBottleCap, BiShadowHandCatchAbreast,
#                  BiShadowHandCatchOver2Underarm, BiShadowHandCatchUnderarm, BiShadowHandDoorOpenInward,
#                  BiShadowHandDoorOpenOutward, BiShadowHandDoorCloseInward, BiShadowHandDoorCloseOutward,
#                  BiShadowHandGraspAndPlace, BiShadowHandLiftUnderarm, BiShadowHandPen, BiShadowHandPointCloud,
#                  BiShadowHandPushBlock, BiShadowHandReOrientation, BiShadowHandScissors, BiShadowHandSwingCup,
#                  BiShadowHandSwitch, BiShadowHandTwoCatchUnderarm
python examples/learning_rl/IsaacGym_RofuncRL/example_DexterousHands_RofuncRL.py --task=BiShadowHandOver --agent=ppo

'''Inference with pre-trained model in model zoo'''
python examples/learning_rl/IsaacGym_RofuncRL/example_DexterousHands_RofuncRL.py --task=BiShadowHandOver --agent=ppo --inference

Task Overview#
Tasks	Animation	Performance	Model Zoo
Ant			✅
Cartpole			✅
FrankaCabinet			✅
FrankaCubeStack			✅
CURICabinet			✅
CURICabinet Image			✅
CURICabinet Bimanual			✅
CURIQbSoftHand SynergyGrasp			✅
Humanoid			✅
HumanoidAMP Backflip			✅
HumanoidAMP Walk
HumanoidAMP Run			✅
HumanoidAMP Dance			✅
HumanoidAMP Hop			✅
HumanoidASE GetupSwordShield			✅
HumanoidASE PerturbSwordShield			✅
HumanoidASE HeadingSwordShield			✅
HumanoidASE ReachSwordShield
HumanoidASE LocationSwordShield			✅
HumanoidASE StrikeSwordShield			✅
BiShadowHand BlockStack			✅
BiShadowHand BottleCap			✅
BiShadowHand CatchAbreast			✅
BiShadowHand CatchOver2Underarm			✅
BiShadowHand CatchUnderarm			✅
BiShadowHand DoorOpenInward			✅
BiShadowHand DoorOpenOutward			✅
BiShadowHand DoorCloseInward			✅
BiShadowHand DoorCloseOutward			✅
BiShadowHand GraspAndPlace			✅
BiShadowHand LiftUnderarm			✅
BiShadowHand Over			✅
BiShadowHand Pen			✅
BiShadowHand PointCloud
BiShadowHand PushBlock			✅
BiShadowHand ReOrientation			✅
BiShadowHand Scissors			✅
BiShadowHand SwingCup			✅
BiShadowHand Switch			✅
BiShadowHand TwoCatchUnderarm			✅

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_AllegroHandOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_AntOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_AnymalOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_AnymalTerrainOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_BallBalanceOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_CartpoleOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_CrazyflieOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_FactoryNutBoltPickOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_FrankaCabinetOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_HumanoidOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_IngenuityOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_QuadcopterOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OmniIsaacGym_RofuncRL/example_ShadowHandOmni_RofuncRL.py --agent=ppo

'''Training'''
python examples/learning_rl/OpenAIGym_RofuncRL/example_GymTasks_RofuncRL.py --task=Gym_Pendulum-v1 --agent=[ppo|a2c|td3|sac]

'''Training'''
python examples/learning_rl/OpenAIGym_RofuncRL/example_GymTasks_RofuncRL.py --task=Gym_CartPole-v1 --agent=[ppo|a2c|td3|sac]

'''Training'''
python examples/learning_rl/OpenAIGym_RofuncRL/example_GymTasks_RofuncRL.py --task=Gym_Acrobot-v1 --agent=[ppo|a2c|td3|sac]

'''Training'''
# Hopper
python examples/learning_rl/D4RL_Rofunc/example_D4RL_RofuncRL.py --task=Hopper --agent=dtrans
# Walker2d
python examples/learning_rl/D4RL_Rofunc/example_D4RL_RofuncRL.py --task=Walker2d --agent=dtrans
# HalfCheetah
python examples/learning_rl/D4RL_Rofunc/example_D4RL_RofuncRL.py --task=HalfCheetah --agent=dtrans
# Reacher2d
python examples/learning_rl/D4RL_Rofunc/example_D4RL_RofuncRL.py --task=Reacher2d --agent=dtrans