context wip

alr_envs/utils/wrapper/dmp_wrapper.py

@@ -98,7 +98,7 @@ class DmpWrapper(MPWrapper):
 
     def mp_rollout(self, action):
         # if self.mp.start_pos is None:
-        self.mp.dmp_start_pos = self.env.init_qpos  # start_pos
+        self.mp.dmp_start_pos = self.env.init_qpos.reshape((1, self.num_dof))  # start_pos
         goal_pos, weight_matrix = self.goal_and_weights(action)
         self.mp.set_weights(weight_matrix, goal_pos)
         return self.mp.reference_trajectory(self.t)

alr_envs/utils/wrapper/mp_wrapper.py

@@ -22,7 +22,7 @@ class MPWrapper(gym.Wrapper, ABC):
                  ):
         super().__init__(env)
 
-        # self.num_dof = num_dof
+        self.num_dof = num_dof
         # self.num_basis = num_basis
         # self.duration = duration  # seconds
 
@@ -50,6 +50,7 @@ class MPWrapper(gym.Wrapper, ABC):
         # for p, c in zip(params, contexts):
         for p in params:
             # self.configure(c)
+            # context = self.reset()
             ob, reward, done, info = self.step(p)
             obs.append(ob)
             rewards.append(reward)