(self)
| 117 | ) |
| 118 | |
| 119 | def reset_model(self) -> npt.NDArray[np.float64]: |
| 120 | self._reset_hand() |
| 121 | self._target_pos = self.goal.copy() |
| 122 | self.obj_init_pos = self.fix_extreme_obj_pos(self.init_config["obj_init_pos"]) |
| 123 | self.obj_init_angle = self.init_config["obj_init_angle"] |
| 124 | |
| 125 | goal_pos = self._get_state_rand_vec() |
| 126 | self._target_pos = goal_pos[3:] |
| 127 | while np.linalg.norm(goal_pos[:2] - self._target_pos[:2]) < 0.15: |
| 128 | goal_pos = self._get_state_rand_vec() |
| 129 | self._target_pos = goal_pos[3:] |
| 130 | self._target_pos = goal_pos[-3:] |
| 131 | self.obj_init_pos = goal_pos[:3] |
| 132 | self._set_obj_xyz(self.obj_init_pos) |
| 133 | |
| 134 | self.model.site("goal").pos = self._target_pos |
| 135 | |
| 136 | self.maxReachDist = np.linalg.norm(self.init_tcp - np.array(self._target_pos)) |
| 137 | |
| 138 | return self._get_obs() |
| 139 | |
| 140 | def compute_reward( |
| 141 | self, actions: npt.NDArray[Any], obs: npt.NDArray[np.float64] |
nothing calls this directly
no test coverage detected