Merge pull request #61 from MuammerBay/feat/reach-ik

Adds reach task with IK controller

Author: louislelay

source/SO_100/SO_100/tasks/reach/__init__.py

@@ -67,3 +67,25 @@
     },
     disable_env_checker=True,
 )
+
+# Relative IK controller
+
+gym.register(
+    id="SO-ARM100-Reach-ROSCON-IK-v0",
+    entry_point="isaaclab.envs:ManagerBasedRLEnv",
+    kwargs={
+        "env_cfg_entry_point": f"{__name__}.ik_rel_env_cfg:SoArm100ReachRosCon_IK_EnvCfg",
+        "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:ReachRosConIKPPORunnerCfg",
+    },
+    disable_env_checker=True,
+)
+
+gym.register(
+    id="SO-ARM100-Reach-ROSCON-IK-Play-v0",
+    entry_point="isaaclab.envs:ManagerBasedRLEnv",
+    kwargs={
+        "env_cfg_entry_point": f"{__name__}.ik_rel_env_cfg:SoArm100ReachRosCon_IK_EnvCfg_PLAY",
+        "rsl_rl_cfg_entry_point": f"{agents.__name__}.rsl_rl_ppo_cfg:ReachRosConIKPPORunnerCfg",
+    },
+    disable_env_checker=True,
+)

source/SO_100/SO_100/tasks/reach/agents/rsl_rl_ppo_cfg.py

@@ -21,7 +21,7 @@ class ReachPPORunnerCfg(RslRlOnPolicyRunnerCfg):
     num_steps_per_env = 24
     max_iterations = 1000
     save_interval = 50
-    experiment_name = "so_arm100_reach"
+    experiment_name = "reach"
     run_name = ""
     resume = False
     empirical_normalization = False
@@ -52,7 +52,38 @@ class ReachRosConPPORunnerCfg(RslRlOnPolicyRunnerCfg):
     num_steps_per_env = 24
     max_iterations = 1000
     save_interval = 50
-    experiment_name = "so_arm100_reach"
+    experiment_name = "reach_roscon"
+    run_name = ""
+    resume = False
+    empirical_normalization = False
+    policy = RslRlPpoActorCriticCfg(
+        init_noise_std=1.0,
+        actor_hidden_dims=[64, 64],
+        critic_hidden_dims=[64, 64],
+        activation="elu",
+    )
+    algorithm = RslRlPpoAlgorithmCfg(
+        value_loss_coef=1.0,
+        use_clipped_value_loss=True,
+        clip_param=0.2,
+        entropy_coef=0.001,
+        num_learning_epochs=8,
+        num_mini_batches=4,
+        learning_rate=1.0e-3,
+        schedule="adaptive",
+        gamma=0.99,
+        lam=0.95,
+        desired_kl=0.01,
+        max_grad_norm=1.0,
+    )
+
+
+@configclass
+class ReachRosConIKPPORunnerCfg(RslRlOnPolicyRunnerCfg):
+    num_steps_per_env = 24
+    max_iterations = 1000
+    save_interval = 50
+    experiment_name = "reach_roscon_ik"
     run_name = ""
     resume = False
     empirical_normalization = False

source/SO_100/SO_100/tasks/reach/ik_rel_env_cfg.py

@@ -0,0 +1,63 @@
+# Copyright (c) 2024-2025, Muammer Bay (LycheeAI), Louis Le Lay
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+#
+# Copyright (c) 2022-2025, The Isaac Lab Project Developers.
+# All rights reserved.
+#
+# SPDX-License-Identifier: BSD-3-Clause
+
+from isaaclab.controllers.differential_ik_cfg import DifferentialIKControllerCfg
+from isaaclab.envs.mdp.actions.actions_cfg import DifferentialInverseKinematicsActionCfg
+from isaaclab.utils import configclass
+
+##
+# Pre-defined configs
+##
+from SO_100.robots import SO_ARM100_CFG, SO_ARM100_ROSCON_CFG  # noqa: F401
+
+from .joint_pos_env_cfg import SoArm100ReachRosConEnvCfg
+
+# ----------------------------------------------------------------
+# --------------- ROSCON ES 2025 asset ---------------------------
+# ----------------------------------------------------------------
+
+
+@configclass
+class SoArm100ReachRosCon_IK_EnvCfg(SoArm100ReachRosConEnvCfg):
+    def __post_init__(self):
+        # post init of parent
+        super().__post_init__()
+
+        # Set Franka as robot
+        # We switch here to a stiffer PD controller for IK tracking to be better.
+        self.scene.robot = SO_ARM100_ROSCON_CFG.replace(prim_path="{ENV_REGEX_NS}/Robot")
+
+        # Set actions for the specific robot type (franka)
+        self.actions.arm_action = DifferentialInverseKinematicsActionCfg(
+            asset_name="robot",
+            joint_names=[
+                "shoulder_pan_joint",
+                "shoulder_lift_joint",
+                "elbow_joint",
+                "wrist_pitch_joint",
+                "wrist_roll_joint",
+            ],
+            body_name="wrist_2_link",
+            controller=DifferentialIKControllerCfg(command_type="pose", use_relative_mode=True, ik_method="dls"),
+            scale=0.5,
+            body_offset=DifferentialInverseKinematicsActionCfg.OffsetCfg(pos=[-0.005, -0.1, 0.0]),
+        )
+
+
+@configclass
+class SoArm100ReachRosCon_IK_EnvCfg_PLAY(SoArm100ReachRosCon_IK_EnvCfg):
+    def __post_init__(self):
+        # post init of parent
+        super().__post_init__()
+        # make a smaller scene for play
+        self.scene.num_envs = 50
+        self.scene.env_spacing = 2.5
+        # disable randomization for play
+        self.observations.policy.enable_corruption = False