mani-skill-nightly 2025.9.14.2244__py3-none-any.whl → 2025.9.14.2255__py3-none-any.whl

mani_skill/examples/motionplanning/base_motionplanner/motionplanner.py

@@ -0,0 +1,195 @@
+import mplib
+import numpy as np
+import sapien
+import trimesh
+
+from mani_skill.agents.base_agent import BaseAgent
+from mani_skill.envs.sapien_env import BaseEnv
+from mani_skill.utils.structs.pose import to_sapien_pose
+
+
+class BaseMotionPlanningSolver:
+
+    def __init__(
+        self,
+        env: BaseEnv,
+        debug: bool = False,
+        vis: bool = True,
+        base_pose: sapien.Pose = None,  # TODO mplib doesn't support robot base being anywhere but 0
+        print_env_info: bool = True,
+        joint_vel_limits=0.9,
+        joint_acc_limits=0.9,
+    ):
+        self.env = env
+        self.base_env: BaseEnv = env.unwrapped
+        self.env_agent: BaseAgent = self.base_env.agent
+        self.robot = self.env_agent.robot
+        self.joint_vel_limits = joint_vel_limits
+        self.joint_acc_limits = joint_acc_limits
+
+        self.base_pose = to_sapien_pose(base_pose)
+
+        self.planner = self.setup_planner()
+        self.control_mode = self.base_env.control_mode
+
+        self.debug = debug
+        self.vis = vis
+        self.print_env_info = print_env_info
+        
+        self.elapsed_steps = 0
+
+        self.use_point_cloud = False
+        self.collision_pts_changed = False
+        self.all_collision_pts = None
+
+    def render_wait(self):
+        if not self.vis or not self.debug:
+            return
+        print("Press [c] to continue")
+        viewer = self.base_env.render_human()
+        while True:
+            if viewer.window.key_down("c"):
+                break
+            self.base_env.render_human()
+
+    def setup_planner(self):
+        move_group = self.MOVE_GROUP if hasattr(self, "MOVE_GROUP") else "eef"
+        link_names = [link.get_name() for link in self.robot.get_links()]
+        joint_names = [joint.get_name() for joint in self.robot.get_active_joints()]
+        planner = mplib.Planner(
+            urdf=self.env_agent.urdf_path,
+            srdf=self.env_agent.urdf_path.replace(".urdf", ".srdf"),
+            user_link_names=link_names,
+            user_joint_names=joint_names,
+            move_group=move_group,
+        )
+        planner.set_base_pose(np.hstack([self.base_pose.p, self.base_pose.q]))
+        planner.joint_vel_limits = np.asarray(planner.joint_vel_limits) * self.joint_vel_limits
+        planner.joint_acc_limits = np.asarray(planner.joint_acc_limits) * self.joint_acc_limits
+        return planner
+
+    def _update_grasp_visual(self, target: sapien.Pose) -> None:
+        return None
+
+    def _transform_pose_for_planning(self, target: sapien.Pose) -> sapien.Pose:
+        return target
+
+    def follow_path(self, result, refine_steps: int = 0):
+        n_step = result["position"].shape[0]
+        for i in range(n_step + refine_steps):
+            qpos = result["position"][min(i, n_step - 1)]
+            if self.control_mode == "pd_joint_pos_vel":
+                qvel = result["velocity"][min(i, n_step - 1)]
+                action = np.hstack([qpos, qvel])
+            else:
+                action = np.hstack([qpos])
+            obs, reward, terminated, truncated, info = self.env.step(action)
+            self.elapsed_steps += 1
+            if self.print_env_info:
+                print(
+                    f"[{self.elapsed_steps:3}] Env Output: reward={reward} info={info}"
+                )
+            if self.vis:
+                self.base_env.render_human()
+        return obs, reward, terminated, truncated, info
+
+        
+    def move_to_pose_with_RRTStar(
+        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
+    ):
+        pose = to_sapien_pose(pose)
+        self._update_grasp_visual(pose)
+        pose = self._transform_pose_for_planning(pose)
+        result = self.planner.plan_qpos_to_pose(
+            np.concatenate([pose.p, pose.q]),
+            self.robot.get_qpos().cpu().numpy()[0],
+            time_step=self.base_env.control_timestep,
+            use_point_cloud=self.use_point_cloud,
+            rrt_range=0.0,
+            planning_time=1,
+            planner_name="RRTstar",
+            wrt_world=True,
+        )
+        if result["status"] != "Success":
+            print(result["status"])
+            self.render_wait()
+            return -1
+        self.render_wait()
+        if dry_run:
+            return result
+        return self.follow_path(result, refine_steps=refine_steps)
+
+    def move_to_pose_with_RRTConnect(
+        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
+    ):
+        pose = to_sapien_pose(pose)
+        self._update_grasp_visual(pose)
+        pose = self._transform_pose_for_planning(pose)
+        result = self.planner.plan_qpos_to_pose(
+            np.concatenate([pose.p, pose.q]),
+            self.robot.get_qpos().cpu().numpy()[0],
+            time_step=self.base_env.control_timestep,
+            use_point_cloud=self.use_point_cloud,
+            wrt_world=True,
+        )
+        if result["status"] != "Success":
+            print(result["status"])
+            self.render_wait()
+            return -1
+        self.render_wait()
+        if dry_run:
+            return result
+        return self.follow_path(result, refine_steps=refine_steps)
+
+    def move_to_pose_with_screw(
+        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
+    ):
+        pose = to_sapien_pose(pose)
+        # try screw two times before giving up
+        self._update_grasp_visual(pose)
+        pose = self._transform_pose_for_planning(pose)
+        result = self.planner.plan_screw(
+            np.concatenate([pose.p, pose.q]),
+            self.robot.get_qpos().cpu().numpy()[0],
+            time_step=self.base_env.control_timestep,
+            use_point_cloud=self.use_point_cloud,
+        )
+        if result["status"] != "Success":
+            result = self.planner.plan_screw(
+                np.concatenate([pose.p, pose.q]),
+                self.robot.get_qpos().cpu().numpy()[0],
+                time_step=self.base_env.control_timestep,
+                use_point_cloud=self.use_point_cloud,
+            )
+            if result["status"] != "Success":
+                print(result["status"])
+                self.render_wait()
+                return -1
+        self.render_wait()
+        if dry_run:
+            return result
+        return self.follow_path(result, refine_steps=refine_steps)
+
+    def add_box_collision(self, extents: np.ndarray, pose: sapien.Pose):
+        self.use_point_cloud = True
+        box = trimesh.creation.box(extents, transform=pose.to_transformation_matrix())
+        pts, _ = trimesh.sample.sample_surface(box, 256)
+        if self.all_collision_pts is None:
+            self.all_collision_pts = pts
+        else:
+            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
+        self.planner.update_point_cloud(self.all_collision_pts)
+
+    def add_collision_pts(self, pts: np.ndarray):
+        if self.all_collision_pts is None:
+            self.all_collision_pts = pts
+        else:
+            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
+        self.planner.update_point_cloud(self.all_collision_pts)
+
+    def clear_collisions(self):
+        self.all_collision_pts = None
+        self.use_point_cloud = False
+
+    def close(self):
+        pass

mani_skill/examples/motionplanning/panda/motionplanner.py

@@ -1,18 +1,16 @@
 import mplib
 import numpy as np
 import sapien
-import trimesh
 
-from mani_skill.agents.base_agent import BaseAgent
 from mani_skill.envs.sapien_env import BaseEnv
-from mani_skill.envs.scene import ManiSkillScene
-from mani_skill.utils.structs.pose import to_sapien_pose
-import sapien.physx as physx
-OPEN = 1
-CLOSED = -1
+from mani_skill.examples.motionplanning.two_finger_gripper.motionplanner import TwoFingerGripperMotionPlanningSolver
 
 
-class PandaArmMotionPlanningSolver:
+class PandaArmMotionPlanningSolver(TwoFingerGripperMotionPlanningSolver):
+    OPEN = 1
+    CLOSED = -1
+    MOVE_GROUP = "panda_hand_tcp"
+
     def __init__(
         self,
         env: BaseEnv,
@@ -24,242 +22,4 @@ class PandaArmMotionPlanningSolver:
         joint_vel_limits=0.9,
         joint_acc_limits=0.9,
     ):
-        self.env = env
-        self.base_env: BaseEnv = env.unwrapped
-        self.env_agent: BaseAgent = self.base_env.agent
-        self.robot = self.env_agent.robot
-        self.joint_vel_limits = joint_vel_limits
-        self.joint_acc_limits = joint_acc_limits
-
-        self.base_pose = to_sapien_pose(base_pose)
-
-        self.planner = self.setup_planner()
-        self.control_mode = self.base_env.control_mode
-
-        self.debug = debug
-        self.vis = vis
-        self.print_env_info = print_env_info
-        self.visualize_target_grasp_pose = visualize_target_grasp_pose
-        self.gripper_state = OPEN
-        self.grasp_pose_visual = None
-        if self.vis and self.visualize_target_grasp_pose:
-            if "grasp_pose_visual" not in self.base_env.scene.actors:
-                self.grasp_pose_visual = build_panda_gripper_grasp_pose_visual(
-                    self.base_env.scene
-                )
-            else:
-                self.grasp_pose_visual = self.base_env.scene.actors["grasp_pose_visual"]
-            self.grasp_pose_visual.set_pose(self.base_env.agent.tcp.pose)
-        self.elapsed_steps = 0
-
-        self.use_point_cloud = False
-        self.collision_pts_changed = False
-        self.all_collision_pts = None
-
-    def render_wait(self):
-        if not self.vis or not self.debug:
-            return
-        print("Press [c] to continue")
-        viewer = self.base_env.render_human()
-        while True:
-            if viewer.window.key_down("c"):
-                break
-            self.base_env.render_human()
-
-    def setup_planner(self):
-        link_names = [link.get_name() for link in self.robot.get_links()]
-        joint_names = [joint.get_name() for joint in self.robot.get_active_joints()]
-        planner = mplib.Planner(
-            urdf=self.env_agent.urdf_path,
-            srdf=self.env_agent.urdf_path.replace(".urdf", ".srdf"),
-            user_link_names=link_names,
-            user_joint_names=joint_names,
-            move_group="panda_hand_tcp",
-            joint_vel_limits=np.ones(7) * self.joint_vel_limits,
-            joint_acc_limits=np.ones(7) * self.joint_acc_limits,
-        )
-        planner.set_base_pose(np.hstack([self.base_pose.p, self.base_pose.q]))
-        return planner
-
-    def follow_path(self, result, refine_steps: int = 0):
-        n_step = result["position"].shape[0]
-        for i in range(n_step + refine_steps):
-            qpos = result["position"][min(i, n_step - 1)]
-            if self.control_mode == "pd_joint_pos_vel":
-                qvel = result["velocity"][min(i, n_step - 1)]
-                action = np.hstack([qpos, qvel, self.gripper_state])
-            else:
-                action = np.hstack([qpos, self.gripper_state])
-            obs, reward, terminated, truncated, info = self.env.step(action)
-            self.elapsed_steps += 1
-            if self.print_env_info:
-                print(
-                    f"[{self.elapsed_steps:3}] Env Output: reward={reward} info={info}"
-                )
-            if self.vis:
-                self.base_env.render_human()
-        return obs, reward, terminated, truncated, info
-
-    def move_to_pose_with_RRTConnect(
-        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
-    ):
-        pose = to_sapien_pose(pose)
-        if self.grasp_pose_visual is not None:
-            self.grasp_pose_visual.set_pose(pose)
-        pose = sapien.Pose(p=pose.p, q=pose.q)
-        result = self.planner.plan_qpos_to_pose(
-            np.concatenate([pose.p, pose.q]),
-            self.robot.get_qpos().cpu().numpy()[0],
-            time_step=self.base_env.control_timestep,
-            use_point_cloud=self.use_point_cloud,
-            wrt_world=True,
-        )
-        if result["status"] != "Success":
-            print(result["status"])
-            self.render_wait()
-            return -1
-        self.render_wait()
-        if dry_run:
-            return result
-        return self.follow_path(result, refine_steps=refine_steps)
-
-    def move_to_pose_with_screw(
-        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
-    ):
-        pose = to_sapien_pose(pose)
-        # try screw two times before giving up
-        if self.grasp_pose_visual is not None:
-            self.grasp_pose_visual.set_pose(pose)
-        pose = sapien.Pose(p=pose.p , q=pose.q)
-        result = self.planner.plan_screw(
-            np.concatenate([pose.p, pose.q]),
-            self.robot.get_qpos().cpu().numpy()[0],
-            time_step=self.base_env.control_timestep,
-            use_point_cloud=self.use_point_cloud,
-        )
-        if result["status"] != "Success":
-            result = self.planner.plan_screw(
-                np.concatenate([pose.p, pose.q]),
-                self.robot.get_qpos().cpu().numpy()[0],
-                time_step=self.base_env.control_timestep,
-                use_point_cloud=self.use_point_cloud,
-            )
-            if result["status"] != "Success":
-                print(result["status"])
-                self.render_wait()
-                return -1
-        self.render_wait()
-        if dry_run:
-            return result
-        return self.follow_path(result, refine_steps=refine_steps)
-
-    def open_gripper(self):
-        self.gripper_state = OPEN
-        qpos = self.robot.get_qpos()[0, :-2].cpu().numpy()
-        for i in range(6):
-            if self.control_mode == "pd_joint_pos":
-                action = np.hstack([qpos, self.gripper_state])
-            else:
-                action = np.hstack([qpos, qpos * 0, self.gripper_state])
-            obs, reward, terminated, truncated, info = self.env.step(action)
-            self.elapsed_steps += 1
-            if self.print_env_info:
-                print(
-                    f"[{self.elapsed_steps:3}] Env Output: reward={reward} info={info}"
-                )
-            if self.vis:
-                self.base_env.render_human()
-        return obs, reward, terminated, truncated, info
-
-    def close_gripper(self, t=6, gripper_state = CLOSED):
-        self.gripper_state = gripper_state
-        qpos = self.robot.get_qpos()[0, :-2].cpu().numpy()
-        for i in range(t):
-            if self.control_mode == "pd_joint_pos":
-                action = np.hstack([qpos, self.gripper_state])
-            else:
-                action = np.hstack([qpos, qpos * 0, self.gripper_state])
-            obs, reward, terminated, truncated, info = self.env.step(action)
-            self.elapsed_steps += 1
-            if self.print_env_info:
-                print(
-                    f"[{self.elapsed_steps:3}] Env Output: reward={reward} info={info}"
-                )
-            if self.vis:
-                self.base_env.render_human()
-        return obs, reward, terminated, truncated, info
-
-    def add_box_collision(self, extents: np.ndarray, pose: sapien.Pose):
-        self.use_point_cloud = True
-        box = trimesh.creation.box(extents, transform=pose.to_transformation_matrix())
-        pts, _ = trimesh.sample.sample_surface(box, 256)
-        if self.all_collision_pts is None:
-            self.all_collision_pts = pts
-        else:
-            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
-        self.planner.update_point_cloud(self.all_collision_pts)
-
-    def add_collision_pts(self, pts: np.ndarray):
-        if self.all_collision_pts is None:
-            self.all_collision_pts = pts
-        else:
-            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
-        self.planner.update_point_cloud(self.all_collision_pts)
-
-    def clear_collisions(self):
-        self.all_collision_pts = None
-        self.use_point_cloud = False
-
-    def close(self):
-        pass
-
-from transforms3d import quaternions
-
-
-def build_panda_gripper_grasp_pose_visual(scene: ManiSkillScene):
-    builder = scene.create_actor_builder()
-    grasp_pose_visual_width = 0.01
-    grasp_width = 0.05
-
-    builder.add_sphere_visual(
-        pose=sapien.Pose(p=[0, 0, 0.0]),
-        radius=grasp_pose_visual_width,
-        material=sapien.render.RenderMaterial(base_color=[0.3, 0.4, 0.8, 0.7])
-    )
-
-    builder.add_box_visual(
-        pose=sapien.Pose(p=[0, 0, -0.08]),
-        half_size=[grasp_pose_visual_width, grasp_pose_visual_width, 0.02],
-        material=sapien.render.RenderMaterial(base_color=[0, 1, 0, 0.7]),
-    )
-    builder.add_box_visual(
-        pose=sapien.Pose(p=[0, 0, -0.05]),
-        half_size=[grasp_pose_visual_width, grasp_width, grasp_pose_visual_width],
-        material=sapien.render.RenderMaterial(base_color=[0, 1, 0, 0.7]),
-    )
-    builder.add_box_visual(
-        pose=sapien.Pose(
-            p=[
-                0.03 - grasp_pose_visual_width * 3,
-                grasp_width + grasp_pose_visual_width,
-                0.03 - 0.05,
-            ],
-            q=quaternions.axangle2quat(np.array([0, 1, 0]), theta=np.pi / 2),
-        ),
-        half_size=[0.04, grasp_pose_visual_width, grasp_pose_visual_width],
-        material=sapien.render.RenderMaterial(base_color=[0, 0, 1, 0.7]),
-    )
-    builder.add_box_visual(
-        pose=sapien.Pose(
-            p=[
-                0.03 - grasp_pose_visual_width * 3,
-                -grasp_width - grasp_pose_visual_width,
-                0.03 - 0.05,
-            ],
-            q=quaternions.axangle2quat(np.array([0, 1, 0]), theta=np.pi / 2),
-        ),
-        half_size=[0.04, grasp_pose_visual_width, grasp_pose_visual_width],
-        material=sapien.render.RenderMaterial(base_color=[1, 0, 0, 0.7]),
-    )
-    grasp_pose_visual = builder.build_kinematic(name="grasp_pose_visual")
-    return grasp_pose_visual
+        super().__init__(env, debug, vis, base_pose, visualize_target_grasp_pose, print_env_info, joint_vel_limits, joint_acc_limits)

mani_skill/examples/motionplanning/panda/motionplanner_stick.py

@@ -1,14 +1,14 @@
 import mplib
 import numpy as np
 import sapien
-import trimesh
-from mani_skill.agents.base_agent import BaseAgent
+
 from mani_skill.envs.sapien_env import BaseEnv
-from mani_skill.envs.scene import ManiSkillScene
-from mani_skill.utils.structs.pose import to_sapien_pose
+from mani_skill.examples.motionplanning.base_motionplanner.motionplanner import BaseMotionPlanningSolver
+
 
+class PandaStickMotionPlanningSolver(BaseMotionPlanningSolver):
+    MOVE_GROUP = "panda_hand_tcp"
 
-class PandaStickMotionPlanningSolver:
     def __init__(
         self,
         env: BaseEnv,
@@ -20,143 +20,4 @@ class PandaStickMotionPlanningSolver:
         joint_vel_limits=0.9,
         joint_acc_limits=0.9,
     ):
-        self.env = env
-        self.base_env: BaseEnv = env.unwrapped
-        self.env_agent: BaseAgent = self.base_env.agent
-        self.robot = self.env_agent.robot
-        self.joint_vel_limits = joint_vel_limits
-        self.joint_acc_limits = joint_acc_limits
-
-        self.base_pose = to_sapien_pose(base_pose)
-
-        self.planner = self.setup_planner()
-        self.control_mode = self.base_env.control_mode
-
-        self.debug = debug
-        self.vis = vis
-        self.print_env_info = print_env_info
-        self.visualize_target_grasp_pose = visualize_target_grasp_pose
-        self.elapsed_steps = 0
-
-        self.use_point_cloud = False
-        self.collision_pts_changed = False
-        self.all_collision_pts = None
-
-    def render_wait(self):
-        if not self.vis or not self.debug:
-            return
-        print("Press [c] to continue")
-        viewer = self.base_env.render_human()
-        while True:
-            if viewer.window.key_down("c"):
-                break
-            self.base_env.render_human()
-
-    def setup_planner(self):
-        link_names = [link.get_name() for link in self.robot.get_links()]
-        joint_names = [joint.get_name() for joint in self.robot.get_active_joints()]
-        planner = mplib.Planner(
-            urdf=self.env_agent.urdf_path,
-            srdf=self.env_agent.urdf_path.replace(".urdf", ".srdf"),
-            user_link_names=link_names,
-            user_joint_names=joint_names,
-            move_group="panda_hand_tcp",
-            joint_vel_limits=np.ones(7) * self.joint_vel_limits,
-            joint_acc_limits=np.ones(7) * self.joint_acc_limits,
-        )
-        planner.set_base_pose(np.hstack([self.base_pose.p, self.base_pose.q]))
-        return planner
-
-    def follow_path(self, result, refine_steps: int = 0):
-        n_step = result["position"].shape[0]
-        for i in range(n_step + refine_steps):
-            qpos = result["position"][min(i, n_step - 1)]
-            if self.control_mode == "pd_joint_pos_vel":
-                qvel = result["velocity"][min(i, n_step - 1)]
-                action = np.hstack([qpos, qvel])
-            else:
-                action = np.hstack([qpos])
-            obs, reward, terminated, truncated, info = self.env.step(action)
-            self.elapsed_steps += 1
-            if self.print_env_info:
-                print(
-                    f"[{self.elapsed_steps:3}] Env Output: reward={reward} info={info}"
-                )
-            if self.vis:
-                self.base_env.render_human()
-        return obs, reward, terminated, truncated, info
-
-    def move_to_pose_with_RRTConnect(
-        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
-    ):
-        pose = to_sapien_pose(pose)
-        if self.grasp_pose_visual is not None:
-            self.grasp_pose_visual.set_pose(pose)
-        pose = sapien.Pose(p=pose.p, q=pose.q)
-        result = self.planner.plan_qpos_to_pose(
-            np.concatenate([pose.p, pose.q]),
-            self.robot.get_qpos().cpu().numpy()[0],
-            time_step=self.base_env.control_timestep,
-            use_point_cloud=self.use_point_cloud,
-            wrt_world=True,
-        )
-        if result["status"] != "Success":
-            print(result["status"])
-            self.render_wait()
-            return -1
-        self.render_wait()
-        if dry_run:
-            return result
-        return self.follow_path(result, refine_steps=refine_steps)
-
-    def move_to_pose_with_screw(
-        self, pose: sapien.Pose, dry_run: bool = False, refine_steps: int = 0
-    ):
-        pose = to_sapien_pose(pose)
-        # try screw two times before giving up
-        pose = sapien.Pose(p=pose.p, q=pose.q)
-        result = self.planner.plan_screw(
-            np.concatenate([pose.p, pose.q]),
-            self.robot.get_qpos().cpu().numpy()[0],
-            time_step=self.base_env.control_timestep,
-            use_point_cloud=self.use_point_cloud,
-        )
-        if result["status"] != "Success":
-            result = self.planner.plan_screw(
-                np.concatenate([pose.p, pose.q]),
-                self.robot.get_qpos().cpu().numpy()[0],
-                time_step=self.base_env.control_timestep,
-                use_point_cloud=self.use_point_cloud,
-            )
-            if result["status"] != "Success":
-                print(result["status"])
-                self.render_wait()
-                return -1
-        self.render_wait()
-        if dry_run:
-            return result
-        return self.follow_path(result, refine_steps=refine_steps)
-
-    def add_box_collision(self, extents: np.ndarray, pose: sapien.Pose):
-        self.use_point_cloud = True
-        box = trimesh.creation.box(extents, transform=pose.to_transformation_matrix())
-        pts, _ = trimesh.sample.sample_surface(box, 256)
-        if self.all_collision_pts is None:
-            self.all_collision_pts = pts
-        else:
-            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
-        self.planner.update_point_cloud(self.all_collision_pts)
-
-    def add_collision_pts(self, pts: np.ndarray):
-        if self.all_collision_pts is None:
-            self.all_collision_pts = pts
-        else:
-            self.all_collision_pts = np.vstack([self.all_collision_pts, pts])
-        self.planner.update_point_cloud(self.all_collision_pts)
-
-    def clear_collisions(self):
-        self.all_collision_pts = None
-        self.use_point_cloud = False
-
-    def close(self):
-        pass
+        super().__init__(env, debug, vis, base_pose, visualize_target_grasp_pose, print_env_info, joint_vel_limits, joint_acc_limits)

mani_skill/examples/motionplanning/panda/solutions/lift_peg_upright.py

@@ -4,7 +4,7 @@ import sapien
 
 from mani_skill.envs.tasks import LiftPegUprightEnv
 from mani_skill.examples.motionplanning.panda.motionplanner import PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import compute_grasp_info_by_obb, get_actor_obb
+from mani_skill.examples.motionplanning.base_motionplanner.utils import compute_grasp_info_by_obb, get_actor_obb
 
 def main():
     env: LiftPegUprightEnv = gym.make(

mani_skill/examples/motionplanning/panda/solutions/peg_insertion_side.py

@@ -5,7 +5,7 @@ import sapien
 from mani_skill.envs.tasks import PegInsertionSideEnv
 from mani_skill.examples.motionplanning.panda.motionplanner import \
     PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import (
+from mani_skill.examples.motionplanning.base_motionplanner.utils import (
     compute_grasp_info_by_obb, get_actor_obb)
 
 

mani_skill/examples/motionplanning/panda/solutions/pick_cube.py

@@ -4,7 +4,7 @@ import sapien
 from mani_skill.envs.tasks import PickCubeEnv
 from mani_skill.examples.motionplanning.panda.motionplanner import \
     PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import (
+from mani_skill.examples.motionplanning.base_motionplanner.utils import (
     compute_grasp_info_by_obb, get_actor_obb)
 
 def solve(env: PickCubeEnv, seed=None, debug=False, vis=False):

mani_skill/examples/motionplanning/panda/solutions/place_sphere.py

@@ -5,7 +5,7 @@ from transforms3d.euler import euler2quat
 from mani_skill.envs.tasks import PlaceSphereEnv
 from mani_skill.utils import common
 from mani_skill.examples.motionplanning.panda.motionplanner import PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import compute_grasp_info_by_obb, get_actor_obb
+from mani_skill.examples.motionplanning.base_motionplanner.utils import compute_grasp_info_by_obb, get_actor_obb
 
 def solve(env: PlaceSphereEnv, seed=None, debug=False, vis=False):
     env.reset(seed=seed)

mani_skill/examples/motionplanning/panda/solutions/plug_charger.py

@@ -8,7 +8,7 @@ from transforms3d.euler import euler2quat
 from mani_skill.envs.tasks import PlugChargerEnv
 from mani_skill.examples.motionplanning.panda.motionplanner import \
     PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import (
+from mani_skill.examples.motionplanning.base_motionplanner.utils import (
     compute_grasp_info_by_obb, get_actor_obb)
 
 

mani_skill/examples/motionplanning/panda/solutions/pull_cube_tool.py

@@ -3,7 +3,7 @@ import sapien
 
 from mani_skill.envs.tasks import PullCubeToolEnv
 from mani_skill.examples.motionplanning.panda.motionplanner import PandaArmMotionPlanningSolver
-from mani_skill.examples.motionplanning.panda.utils import compute_grasp_info_by_obb, get_actor_obb
+from mani_skill.examples.motionplanning.base_motionplanner.utils import compute_grasp_info_by_obb, get_actor_obb
 
 def solve(env: PullCubeToolEnv, seed=None, debug=False, vis=False):
     env.reset(seed=seed)