updating rendering of goal positions

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_basketball_v2.py

@@ -105,6 +105,7 @@ def reset_model(self):
             mujoco.mj_name2id(self.model, mujoco.mjtObj.mjOBJ_SITE, "goal")
         ]
         self._set_obj_xyz(self.obj_init_pos)
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_box_close_v2.py

@@ -107,7 +107,7 @@ def reset_model(self):
             mujoco.mj_step(self.model, self.data)
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     @staticmethod

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_coffee_pull_v2.py

@@ -109,6 +109,7 @@ def reset_model(self):
         ] = pos_machine
 
         self._target_pos = pos_mug_goal
+        self.model.site('mug_goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_coffee_push_v2.py

@@ -110,6 +110,7 @@ def reset_model(self):
         ] = pos_machine
 
         self._target_pos = pos_mug_goal
+        self.model.site('coffee_goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_dial_turn_v2.py

@@ -101,6 +101,7 @@ def reset_model(self):
             mujoco.mj_name2id(self.model, mujoco.mjtObj.mjOBJ_BODY, "dial")
         ] = self.obj_init_pos
         self.dial_push_position = self._get_pos_objects() + np.array([0.05, 0.02, 0.09])
+        self.model.site('goal').pos = self._target_pos
         mujoco.mj_forward(self.model, self.data)
         return self._get_obs()
 

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_door_close_v2.py

@@ -80,7 +80,7 @@ def reset_model(self):
 
         # keep the door open after resetting initial positions
         self._set_obj_xyz(-1.5708)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     @_assert_task_is_set

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_door_v2.py

@@ -108,7 +108,7 @@ def reset_model(self):
             self.data.geom("handle").xpos[:-1] - self._target_pos[:-1]
         )
         self.target_reward = 1000 * self.maxPullDist + 1000 * 2
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     @staticmethod

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_drawer_close_v2.py

@@ -108,7 +108,7 @@ def reset_model(self):
         # Pull drawer out all the way and mark its starting position
         self._set_obj_xyz(-self.maxDist)
         self.obj_init_pos = self._get_pos_objects()
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_drawer_open_v2.py

@@ -103,8 +103,7 @@ def reset_model(self):
         self._target_pos = self.obj_init_pos + np.array(
             [0.0, -0.16 - self.maxDist, 0.09]
         )
-        mujoco.mj_forward(self.model, self.data)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_faucet_close_v2.py

@@ -98,6 +98,7 @@ def reset_model(self):
             [-self._handle_length, 0.0, 0.125]
         )
         mujoco.mj_forward(self.model, self.data)
+        self.model.site('goal_close').pos = self._target_pos
         return self._get_obs()
 
     def _reset_hand(self):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_faucet_open_v2.py

@@ -97,7 +97,7 @@ def reset_model(self):
         self._target_pos = self.obj_init_pos + np.array(
             [+self._handle_length, 0.0, 0.125]
         )
-        mujoco.mj_forward(self.model, self.data)
+        self.model.site('goal_open').pos = self._target_pos
         return self._get_obs()
 
     def _reset_hand(self):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_hammer_v2.py

@@ -104,7 +104,7 @@ def reset_model(self):
         self.nail_init_pos = self._get_site_pos("nailHead")
         self.obj_init_pos = self.hammer_init_pos.copy()
         self._set_hammer_xyz(self.hammer_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     @staticmethod

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_hand_insert_v2.py

@@ -101,6 +101,7 @@ def reset_model(self):
         self._target_pos = goal_pos[-3:]
 
         self._set_obj_xyz(self.obj_init_pos)
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_handle_press_side_v2.py

@@ -74,7 +74,7 @@ def evaluate_state(self, obs, action):
             object_grasped,
             in_place,
         ) = self.compute_reward(action, obs)
-
+        print(target_to_obj)
         info = {
             "success": float(target_to_obj <= self.TARGET_RADIUS),
             "near_object": float(tcp_to_obj <= 0.05),

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_lever_pull_v2.py

@@ -108,7 +108,7 @@ def reset_model(self):
         self._target_pos = self.obj_init_pos + np.array(
             [0.12, 0.0, 0.25 + self.LEVER_RADIUS]
         )
-        mujoco.mj_forward(self.model, self.data)
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_peg_insertion_side_v2.py

@@ -127,6 +127,7 @@ def reset_model(self):
             mujoco.mj_name2id(self.model, mujoco.mjtObj.mjOBJ_BODY, "box")
         ] = pos_box
         self._target_pos = pos_box + np.array([0.03, 0.0, 0.13])
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_peg_unplug_side_v2.py

@@ -100,7 +100,7 @@ def reset_model(self):
         self.obj_init_pos = self._get_site_pos("pegEnd")
 
         self._target_pos = pos_plug + np.array([0.15, 0.0, 0.0])
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_pick_out_of_hole_v2.py

@@ -103,7 +103,7 @@ def reset_model(self):
         self.obj_init_pos = pos_obj
         self._set_obj_xyz(self.obj_init_pos)
         self._target_pos = pos_goal
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_pick_place_v2.py

@@ -138,7 +138,7 @@ def reset_model(self):
         self.init_right_pad = self.get_body_com("rightpad")
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def _gripper_caging_reward(self, action, obj_position):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_pick_place_wall_v2.py

@@ -130,7 +130,7 @@ def reset_model(self):
         self.obj_init_pos = goal_pos[:3]
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_push_back_v2.py

@@ -114,7 +114,7 @@ def reset_model(self):
         self.obj_init_pos = np.concatenate((goal_pos[:2], [self.obj_init_pos[-1]]))
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def _gripper_caging_reward(self, action, obj_position, obj_radius):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_push_v2.py

@@ -135,7 +135,7 @@ def reset_model(self):
         self.obj_init_pos = np.concatenate((goal_pos[:2], [self.obj_init_pos[-1]]))
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_push_wall_v2.py

@@ -128,6 +128,7 @@ def reset_model(self):
         self.obj_init_pos = np.concatenate((goal_pos[:2], [self.obj_init_pos[-1]]))
 
         self._set_obj_xyz(self.obj_init_pos)
+        self.model.site('goal').pos = self._target_pos
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_reach_v2.py

@@ -113,7 +113,8 @@ def reset_model(self):
         self._target_pos = goal_pos[-3:]
         self.obj_init_pos = goal_pos[:3]
         self._set_obj_xyz(self.obj_init_pos)
-        mujoco.mj_forward(self.model, self.data)
+
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def compute_reward(self, actions, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_reach_wall_v2.py

@@ -104,7 +104,7 @@ def reset_model(self):
         self.obj_init_pos = goal_pos[:3]
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def compute_reward(self, actions, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_shelf_place_v2.py

@@ -125,7 +125,7 @@ def reset_model(self):
         )
 
         self._set_obj_xyz(self.obj_init_pos)
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def compute_reward(self, action, obs):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_soccer_v2.py

@@ -108,7 +108,7 @@ def reset_model(self):
         self.maxPushDist = np.linalg.norm(
             self.obj_init_pos[:2] - np.array(self._target_pos)[:2]
         )
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _gripper_caging_reward(self, action, obj_position, obj_radius):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_stick_pull_v2.py

@@ -146,7 +146,7 @@ def reset_model(self):
         self._set_stick_xyz(self.stick_init_pos)
         self._set_obj_xyz(self.obj_init_qpos)
         self.obj_init_pos = self.get_body_com("object").copy()
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _stick_is_inserted(self, handle, end_of_stick):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_stick_push_v2.py

@@ -143,7 +143,7 @@ def reset_model(self):
         self._set_stick_xyz(self.stick_init_pos)
         self._set_obj_xyz(self.obj_init_qpos)
         self.obj_init_pos = self.get_body_com("object").copy()
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _gripper_caging_reward(

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_sweep_into_goal_v2.py

@@ -98,7 +98,7 @@ def reset_model(self):
         self.maxPushDist = np.linalg.norm(
             self.obj_init_pos[:2] - np.array(self._target_pos)[:2]
         )
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _gripper_caging_reward(self, action, obj_position, obj_radius):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_sweep_v2.py

@@ -97,7 +97,7 @@ def reset_model(self):
             self.get_body_com("obj")[:-1] - self._target_pos[:-1]
         )
         self.target_reward = 1000 * self.maxPushDist + 1000 * 2
-
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _gripper_caging_reward(self, action, obj_position, obj_radius):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_window_close_v2.py

@@ -112,7 +112,7 @@ def reset_model(self):
             [0.2, 0.0, 0.0]
         )
         self.data.joint("window_slide").qpos = 0.2
-        mujoco.mj_forward(self.model, self.data)
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def _reset_hand(self):

metaworld/envs/mujoco/sawyer_xyz/v2/sawyer_window_open_v2.py

@@ -111,7 +111,7 @@ def reset_model(self):
 
         self.window_handle_pos_init = self._get_pos_objects()
         self.data.joint("window_slide").qpos = 0.0
-        mujoco.mj_forward(self.model, self.data)
+        self._set_pos_site('goal', self._target_pos)
         return self._get_obs()
 
     def compute_reward(self, actions, obs):