Merge pull request abizovnuralem#13 from abizovnuralem/added_g1

removed the hardcode and add G1 support

README.md

@@ -12,6 +12,13 @@ Real time Go2 Balancing:
 </p>
 
 
+Real time G1 Balancing:
+
+<p align="center">
+<img width="1280" height="600" src="https://github.com/abizovnuralem/go2_omniverse/assets/33475993/d035f72a-8996-461c-a902-38e68052d029" alt='Go2'>
+</p>
+
+
 Go2 Ros2 Camera stream:
 
 <p align="center">
@@ -72,6 +79,7 @@ VR support:
 9. Nav2 with Slam_toolbox :white_check_mark:
 10. Bunch of RL-envs for custom dog training :white_check_mark:
 11. Custom numbers of robots :white_check_mark:
+12. Added G1 Unitree support :white_check_mark:
 
 ## Your feedback and support mean the world to us. 
 
@@ -127,12 +135,18 @@ with Isaac_sim/extension.toml in this repo.
 
 ## Usage
 The current project was tested on Ubuntu 22.04, IsaacSim 4.0 with Isaac Lab and Nvidia Driver Version: 545.
-To start the project, execute:
+To start the project with Unitree GO2, execute:
 
 ```
 ./run_sim.sh
 ```
 
+To start the project with Unitree G1, execute:
+
+```
+./run_sim_g1.sh
+```
+
 You can control the dog using "WASD" keyboard commands
 
 ## ROS2 SDK

agent_cfg.py

@@ -59,4 +59,44 @@
         'resume': False, 
         'load_run': '.*', 
         'load_checkpoint': 'model_.*.pt'
+        }
+
+
+unitree_g1_agent_cfg = {
+        'seed': 42, 
+        'device': 'cuda', 
+        'num_steps_per_env': 24, 
+        'max_iterations': 15000, 
+        'empirical_normalization': False, 
+        'policy': {
+            'class_name': 'ActorCritic', 
+            'init_noise_std': 1.0, 
+            'actor_hidden_dims': [512, 256, 128], 
+            'critic_hidden_dims': [512, 256, 128], 
+            'activation': 'elu'
+            }, 
+        'algorithm': {
+            'class_name': 'PPO', 
+            'value_loss_coef': 1.0, 
+            'use_clipped_value_loss': True, 
+            'clip_param': 0.2, 
+            'entropy_coef': 0.01, 
+            'num_learning_epochs': 5, 
+            'num_mini_batches': 4, 
+            'learning_rate': 0.001, 
+            'schedule': 'adaptive', 
+            'gamma': 0.99, 
+            'lam': 0.95, 
+            'desired_kl': 0.01, 
+            'max_grad_norm': 1.0
+        }, 
+        'save_interval': 50, 
+        'experiment_name': 'g1_rough', 
+        'run_name': '', 
+        'logger': 'tensorboard', 
+        'neptune_project': 'orbit', 
+        'wandb_project': 'orbit', 
+        'resume': False, 
+        'load_run': '.*', 
+        'load_checkpoint': 'model_.*.pt'
         }

custom_rl_env.py

@@ -49,6 +49,9 @@
 import omni.isaac.lab_tasks.manager_based.locomotion.velocity.mdp as mdp
 
 
+from robots.g1.config import G1_CFG
+
+
 base_command = []
 
 
@@ -298,4 +301,23 @@ def __post_init__(self):
         self.rewards.dof_acc_l2.weight = -2.5e-7
 
         # terminations
-        self.terminations.base_contact.params["sensor_cfg"].body_names = "base"
+        self.terminations.base_contact.params["sensor_cfg"].body_names = "base"
+
+
+@configclass
+class G1RoughEnvCfg(LocomotionVelocityRoughEnvCfg):
+    def __post_init__(self):
+        # post init of parent
+        super().__post_init__()
+        # Scene
+        G1_MINIMAL_CFG = G1_CFG.copy()
+        G1_MINIMAL_CFG.spawn.usd_path = "./robots/g1/g1.usd"
+        self.scene.robot = G1_MINIMAL_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
+        self.scene.height_scanner.prim_path = "{ENV_REGEX_NS}/Robot/torso_link"
+
+        # rewards
+        self.rewards.feet_air_time.params["sensor_cfg"].body_names = ".*_ankle_roll_link"
+        self.rewards.undesired_contacts = None
+
+        # Terminations
+        self.terminations.base_contact.params["sensor_cfg"].body_names = ["torso_link"]

logs/rsl_rl/g1_rough/2024-06-03_21-09-07/git/IsaacLab.diff

@@ -0,0 +1,241 @@
+--- git status ---
+On branch main
+Your branch is up to date with 'origin/main'.
+
+Changes not staged for commit:
+  (use "git add <file>..." to update what will be committed)
+  (use "git restore <file>..." to discard changes in working directory)
+	modified:   source/extensions/omni.isaac.lab_assets/omni/isaac/lab_assets/unitree.py
+	modified:   source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/mdp/rewards.py
+	modified:   source/standalone/workflows/rsl_rl/train.py
+
+Untracked files:
+  (use "git add <file>..." to include in what will be committed)
+	source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/config/g1/
+	source/exts/
+
+no changes added to commit (use "git add" and/or "git commit -a") 
+
+
+--- git diff ---
+diff --git a/source/extensions/omni.isaac.lab_assets/omni/isaac/lab_assets/unitree.py b/source/extensions/omni.isaac.lab_assets/omni/isaac/lab_assets/unitree.py
+index 8533010..43f55b2 100644
+--- a/source/extensions/omni.isaac.lab_assets/omni/isaac/lab_assets/unitree.py
++++ b/source/extensions/omni.isaac.lab_assets/omni/isaac/lab_assets/unitree.py
+@@ -265,3 +265,160 @@ H1_MINIMAL_CFG.spawn.usd_path = f"{ISAACLAB_NUCLEUS_DIR}/Robots/Unitree/H1/h1_mi
+
+ This configuration removes most collision meshes to speed up simulation.
+ """
++
++
++
++G1_CFG = ArticulationCfg(
++    spawn=sim_utils.UsdFileCfg(
++        usd_path="/home/brimo/Desktop/g1/g1_t.usd",
++        activate_contact_sensors=True,
++        rigid_props=sim_utils.RigidBodyPropertiesCfg(
++            disable_gravity=False,
++            retain_accelerations=False,
++            linear_damping=0.0,
++            angular_damping=0.0,
++            max_linear_velocity=1000.0,
++            max_angular_velocity=1000.0,
++            max_depenetration_velocity=1.0,
++        ),
++        articulation_props=sim_utils.ArticulationRootPropertiesCfg(
++            enabled_self_collisions=False, solver_position_iteration_count=4, solver_velocity_iteration_count=4
++        ),
++    ),
++    init_state=ArticulationCfg.InitialStateCfg(
++        pos=(0.0, 0.0, 0.70),
++        joint_pos={
++            
++            ".*_hip_pitch_joint": -0.28,
++            ".*_knee_joint": 0.63,
++            ".*_ankle_pitch_joint": -0.35,
++
++            ".*_elbow_pitch_joint": 0.87,
++            "left_shoulder_roll_joint": 0.16,
++            "left_shoulder_pitch_joint": 0.35,
++            
++            "right_shoulder_roll_joint": -0.16,
++            "right_shoulder_pitch_joint": 0.35,
++            
++            "left_one_joint": 1.0,
++            "right_one_joint": -1.0,
++
++
++        },
++        joint_vel={".*": 0.0},
++    ),
++    soft_joint_pos_limit_factor=0.9,
++    actuators={
++        "legs": ImplicitActuatorCfg(
++            joint_names_expr=[
++                ".*_hip_yaw_joint", 
++                ".*_hip_roll_joint",
++                 ".*_hip_pitch_joint", 
++                ".*_knee_joint",
++                "torso_joint"
++            ],
++            effort_limit=300,
++            velocity_limit=100.0,
++            stiffness={
++                ".*_hip_yaw_joint": 150.0,
++                ".*_hip_roll_joint": 150.0,
++                ".*_hip_pitch_joint": 200.0,
++                ".*_knee_joint": 200.0,
++                "torso_joint": 200.0,
++            },
++            damping={
++                ".*_hip_yaw_joint": 5.0,
++                ".*_hip_roll_joint": 5.0,
++                ".*_hip_pitch_joint": 5.0,
++                ".*_knee_joint": 5.0,
++                "torso_joint": 5.0,
++            },
++            armature={
++                ".*_hip_yaw_joint": 0.1,
++                ".*_hip_roll_joint": 0.1,
++                ".*_hip_pitch_joint": 0.1,
++                ".*_knee_joint": 0.1,
++                "torso_joint": 0.1,
++            },
++        ),
++        "feet": ImplicitActuatorCfg(
++            joint_names_expr=[
++                ".*_ankle_pitch_joint", 
++                ".*_ankle_roll_joint"
++            ],
++            stiffness={
++                ".*_ankle_pitch_joint": 20.0,
++                ".*_ankle_roll_joint": 20.0
++            },
++            damping={
++                ".*_ankle_pitch_joint": 4.0,
++                ".*_ankle_roll_joint": 4.0
++            },
++            armature={
++                ".*_ankle_pitch_joint": 0.1,
++                ".*_ankle_roll_joint": 0.1
++            },
++        ),
++        "arms": ImplicitActuatorCfg(
++            joint_names_expr=[
++                ".*_shoulder_pitch_joint",
++                ".*_shoulder_roll_joint",
++                ".*_shoulder_yaw_joint",
++                ".*_elbow_pitch_joint",
++                ".*_elbow_roll_joint",
++                ".*_five_joint",
++                ".*_three_joint",
++                ".*_six_joint",
++                ".*_four_joint",
++                ".*_zero_joint",
++                ".*_one_joint",
++                ".*_two_joint"
++            ],
++            effort_limit=300,
++            velocity_limit=100.0,
++            stiffness={
++                ".*_shoulder_pitch_joint": 40.0,
++                ".*_shoulder_roll_joint": 40.0,
++                ".*_shoulder_yaw_joint": 40.0,
++                ".*_elbow_pitch_joint": 40.0,
++                ".*_elbow_roll_joint": 40.0,
++                ".*_five_joint": 40.0,
++                ".*_three_joint": 40.0,
++                ".*_six_joint": 40.0,
++                ".*_four_joint": 40.0,
++                ".*_zero_joint": 40.0,
++                ".*_one_joint": 40.0,
++                ".*_two_joint": 40.0,
++            },
++            damping={
++                ".*_shoulder_pitch_joint": 10.0,
++                ".*_shoulder_roll_joint": 10.0,
++                ".*_shoulder_yaw_joint": 10.0,
++                ".*_elbow_pitch_joint": 10.0,
++                ".*_elbow_roll_joint": 10.0,
++                ".*_five_joint": 10.0,
++                ".*_three_joint": 10.0,
++                ".*_six_joint": 10.0,
++                ".*_four_joint": 10.0,
++                ".*_zero_joint": 10.0,
++                ".*_one_joint": 10.0,
++                ".*_two_joint": 10.0,
++            },
++            armature={
++                ".*_shoulder_pitch_joint": 0.1,
++                ".*_shoulder_roll_joint": 0.1,
++                ".*_shoulder_yaw_joint": 0.1,
++                ".*_elbow_pitch_joint": 0.1,
++                ".*_elbow_roll_joint": 0.1,
++                ".*_five_joint": 0.1,
++                ".*_three_joint": 0.1,
++                ".*_six_joint": 0.1,
++                ".*_four_joint": 0.1,
++                ".*_zero_joint": 0.1,
++                ".*_one_joint": 0.1,
++                ".*_two_joint": 0.1,
++            },
++        ),
++    },
++)
++"""Configuration for the Unitree G1 Humanoid robot."""
+diff --git a/source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/mdp/rewards.py b/source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/mdp/rewards.py
+index b3a1756..96d2612 100644
+--- a/source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/mdp/rewards.py
++++ b/source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/manager_based/locomotion/velocity/mdp/rewards.py
+@@ -47,6 +47,7 @@ def feet_air_time_positive_biped(env, command_name: str, threshold: float, senso
+     If the commands are small (i.e. the agent is not supposed to take a step), then the reward is zero.
+     """
+     contact_sensor: ContactSensor = env.scene.sensors[sensor_cfg.name]
++
+     # compute the reward
+     air_time = contact_sensor.data.current_air_time[:, sensor_cfg.body_ids]
+     contact_time = contact_sensor.data.current_contact_time[:, sensor_cfg.body_ids]
+@@ -60,6 +61,28 @@ def feet_air_time_positive_biped(env, command_name: str, threshold: float, senso
+     return reward
+
+
++def feet_leave_ground(env, sensor_cfg: SceneEntityCfg, threshold: float) -> torch.Tensor:
++    """
++    Terminate if the robot's feet have left the ground.
++
++    Args:
++        env: The environment containing the sensors.
++        threshold (float): The threshold below which a contact force is considered as the feet having left the ground.
++        sensor_cfg (SceneEntityCfg): Configuration for the contact sensor.
++
++    Returns:
++        torch.Tensor: A tensor indicating whether the feet have left the ground for each instance.
++    """
++    contact_sensor = env.scene.sensors[sensor_cfg.name]
++    contact_forces = contact_sensor.data.net_forces_w[:, sensor_cfg.body_ids].norm(dim=-1)
++
++    # Check if contact forces are below the threshold
++    feet_off_ground = torch.all(contact_forces < threshold, dim=1)
++
++    return feet_off_ground
++
++
++
+ def feet_slide(env, sensor_cfg: SceneEntityCfg, asset_cfg: SceneEntityCfg = SceneEntityCfg("robot")) -> torch.Tensor:
+     # Penalize feet sliding
+     contact_sensor: ContactSensor = env.scene.sensors[sensor_cfg.name]
+diff --git a/source/standalone/workflows/rsl_rl/train.py b/source/standalone/workflows/rsl_rl/train.py
+index 748d9bd..d351b2a 100644
+--- a/source/standalone/workflows/rsl_rl/train.py
++++ b/source/standalone/workflows/rsl_rl/train.py
+@@ -24,8 +24,8 @@ parser.add_argument("--cpu", action="store_true", default=False, help="Use CPU p
+ parser.add_argument(
+     "--disable_fabric", action="store_true", default=False, help="Disable fabric and use USD I/O operations."
+ )
+-parser.add_argument("--num_envs", type=int, default=None, help="Number of environments to simulate.")
+-parser.add_argument("--task", type=str, default=None, help="Name of the task.")
++parser.add_argument("--num_envs", type=int, default=1, help="Number of environments to simulate.")
++parser.add_argument("--task", type=str, default="Isaac-Velocity-Rough-G1-v0", help="Name of the task.")
+ parser.add_argument("--seed", type=int, default=None, help="Seed used for the environment")
+ parser.add_argument("--max_iterations", type=int, default=None, help="RL Policy training iterations.")
+ # append RSL-RL cli arguments