import os from copy import deepcopy import numpy as np from core.skills.base_skill import BaseSkill, register_skill from core.utils.usd_geom_utils import compute_bbox from omegaconf import DictConfig, OmegaConf from omni.isaac.core.controllers import BaseController from omni.isaac.core.robots.robot import Robot from omni.isaac.core.tasks import BaseTask from omni.isaac.core.utils.prims import get_prim_at_path from omni.isaac.core.utils.transformations import ( get_relative_transform, pose_from_tf_matrix, tf_matrix_from_pose, ) from scipy.spatial.transform import Rotation as R from scipy.spatial.transform import Slerp # pylint: disable=unused-argument @register_skill class Dexplace(BaseSkill): def __init__(self, robot: Robot, controller: BaseController, task: BaseTask, cfg: DictConfig, *args, **kwargs): super().__init__() self.robot = robot self.controller = controller self.task = task self.skill_cfg = cfg self.name = cfg["name"] self.pick_obj = task._task_objects[cfg["objects"][0]] self.place_obj = task._task_objects[cfg["objects"][1]] self.gripper_axis = cfg.get("gripper_axis", None) self.camera_axis_filter = cfg.get("camera_axis_filter", None) self.place_part_prim_path = cfg.get("place_part_prim_path", None) # Get place annotation usd_path = [obj["path"] for obj in task.cfg["objects"] if obj["name"] == self.skill_cfg["objects"][0]][0] usd_path = os.path.join(self.task.asset_root, usd_path) place_range_path = usd_path.replace("Aligned_obj.usd", "place_range.yaml") if os.path.exists(place_range_path): with open(place_range_path, "r", encoding="utf-8") as f: place_data = OmegaConf.load(f) self.x_range = place_data.x_range self.y_range = place_data.y_range else: self.x_range = [0.4, 0.6] self.y_range = [0.4, 0.6] # Get place_prim if self.place_part_prim_path: self.place_prim_path = f"{self.place_obj.prim_path}/{self.place_part_prim_path}" else: self.place_prim_path = self.place_obj.prim_path # Get left or right if "left" in self.controller.robot_file: self.robot_ee_path = self.robot.fl_ee_path self.robot_base_path = self.robot.fl_base_path elif "right" in self.controller.robot_file: self.robot_ee_path = self.robot.fr_ee_path self.robot_base_path = self.robot.fr_base_path if kwargs: self.draw = kwargs["draw"] self.manip_list = [] def simple_generate_manip_cmds(self): manip_list = [] place_traj, post_place_level = self.sample_gripper_place_traj() if len(place_traj) > 1: # Having waypoints for waypoint in place_traj[:-1]: p_base_ee_mid, q_base_ee_mid = waypoint[:3], waypoint[3:] cmd = (p_base_ee_mid, q_base_ee_mid, "close_gripper", {}) manip_list.append(cmd) # The last waypoint p_base_ee_place, q_base_ee_place = place_traj[-1][:3], place_traj[-1][3:] cmd = (p_base_ee_place, q_base_ee_place, "close_gripper", {}) manip_list.append(cmd) cmd = (p_base_ee_place, q_base_ee_place, "open_gripper", {}) manip_list.extend([cmd] * self.skill_cfg.get("gripper_change_steps", 10)) # Adding a pose place pose to avoid collision when combining place skill and close skill T_base_ee_place = tf_matrix_from_pose(p_base_ee_place, q_base_ee_place) # Post place T_base_ee_postplace = deepcopy(T_base_ee_place) # Retreat for a bit along gripper axis if "r5a" in self.controller.robot_file: T_base_ee_postplace[0:3, 3] = T_base_ee_postplace[0:3, 3] - T_base_ee_postplace[0:3, 0] * post_place_level else: T_base_ee_postplace[0:3, 3] = T_base_ee_postplace[0:3, 3] - T_base_ee_postplace[0:3, 2] * post_place_level cmd = (*pose_from_tf_matrix(T_base_ee_postplace), "open_gripper", {}) manip_list.append(cmd) self.manip_list = manip_list def sample_gripper_place_traj(self): place_traj = [] T_base_ee = get_relative_transform(get_prim_at_path(self.robot_ee_path), get_prim_at_path(self.robot_base_path)) T_world_base = get_relative_transform( get_prim_at_path(self.robot_base_path), get_prim_at_path(self.task.root_prim_path) ) T_world_ee = T_world_base @ T_base_ee p_world_ee_start, q_world_ee_start = pose_from_tf_matrix(T_world_ee) # Getting the object pose T_world_obj = tf_matrix_from_pose(*self.pick_obj.get_local_pose()) # Calculate the pose of the end-effector in the object's coordinate frame T_obj_world = np.linalg.inv(T_world_obj) # Getting the relation pose and distance of ee to object (after picking, before placing) T_obj_ee = T_obj_world @ T_world_ee ee2o_distance = np.linalg.norm(T_obj_ee[0:3, 3]) place_part_prim = get_prim_at_path(self.place_prim_path) bbox_place_obj = compute_bbox(place_part_prim) x_min, y_min, z_min = bbox_place_obj.min x_max, y_max, z_max = bbox_place_obj.max self.place_boundary = [[x_min, y_min, z_min], [x_max, y_max, z_max]] # Calculate the bounding box vertices vertices = [ [x_min, y_min, z_min], [x_min, y_max, z_min], [x_max, y_min, z_min], [x_max, y_max, z_min], [x_min, y_min, z_max], [x_min, y_max, z_max], [x_max, y_min, z_max], [x_max, y_max, z_max], ] # Draw the bounding box vertices if self.draw: for vertex in vertices: self.draw.draw_points([vertex], [(0, 0, 0, 1)], [7]) # black # 1. Obtaining ee_ori p_world_ee_init = self.controller.T_world_ee_init[0:3, 3] # getting initial ee position container_position = self.place_obj.get_local_pose()[0] # getting container position container_position[1] += 0.0 gripper_axis = container_position - p_world_ee_init # gripper_axis is aligned with the container direction gripper_axis = gripper_axis / np.linalg.norm(gripper_axis) # Normalize the target vector q_world_ee = self.get_ee_ori(gripper_axis, T_world_ee, self.camera_axis_filter) # 2. Obtaining p_world_ee x = x_min + np.random.uniform(self.x_range[0], self.x_range[1]) * (x_max - x_min) y = y_min + np.random.uniform(self.y_range[0], self.y_range[1]) * (y_max - y_min) z = z_min + 0.15 obj_place_position = np.array([x, y, z]) if self.draw: self.draw.draw_points([obj_place_position.tolist()], [(1, 0, 0, 1)], [7]) # red p_world_ee = obj_place_position - gripper_axis * ee2o_distance # 3. Adding Waypoint # Pre place p_world_ee_mid = (p_world_ee_start + p_world_ee) / 2.0 p_world_ee_mid[2] += 0.05 slerp = Slerp([0, 1], R.from_quat([q_world_ee_start, q_world_ee])) q_world_ee_mid = slerp([0.5]).as_quat()[0] if self.draw: self.draw.draw_points([p_world_ee_mid.tolist()], [(0, 1, 0, 1)], [7]) # green place_traj.append(self.adding_waypoint(p_world_ee_mid, q_world_ee_mid, T_world_base)) # Place if self.draw: self.draw.draw_points([p_world_ee.tolist()], [(0, 1, 0, 1)], [7]) # green place_traj.append(self.adding_waypoint(p_world_ee, q_world_ee, T_world_base)) post_place_level = 0.1 return place_traj, post_place_level def get_ee_ori(self, gripper_axis, T_world_ee, camera_axis_filter=None): gripper_x = gripper_axis if camera_axis_filter is not None: direction = camera_axis_filter[0]["direction"] degree = camera_axis_filter[1]["degree"] direction = np.array(direction) / np.linalg.norm(direction) # Normalize the direction vector angle = np.radians(np.random.uniform(degree[0], degree[1])) gripper_z = direction - np.dot(direction, gripper_x) * gripper_x gripper_z = gripper_z / np.linalg.norm(gripper_z) rotation_axis = np.cross(gripper_z, gripper_x) rotation_axis = rotation_axis / np.linalg.norm(rotation_axis) gripper_z = R.from_rotvec(angle * rotation_axis).apply(gripper_z) else: current_z = T_world_ee[0:3, 1] gripper_z = current_z - np.dot(current_z, gripper_x) * gripper_x gripper_z = gripper_z / np.linalg.norm(gripper_z) gripper_y = np.cross(gripper_z, gripper_x) gripper_y = gripper_y / np.linalg.norm(gripper_y) gripper_z = np.cross(gripper_x, gripper_y) R_world_ee = np.column_stack((gripper_x, gripper_y, gripper_z)) q_world_ee = R.from_matrix(R_world_ee).as_quat(scalar_first=True) return q_world_ee def adding_waypoint(self, p_world_ee, q_world_ee, T_world_base): """ Adding a waypoint, also transform from wolrd frame to robot frame """ T_world_ee = tf_matrix_from_pose(p_world_ee, q_world_ee) T_base_ee = np.linalg.inv(T_world_base) @ T_world_ee p_base_ee, q_base_ee = pose_from_tf_matrix(T_base_ee) waypoint = np.concatenate((p_base_ee, q_base_ee)) return waypoint def is_feasible(self, th=10): return self.controller.num_plan_failed <= th def is_subtask_done(self, t_eps=1e-3, o_eps=5e-3): assert len(self.manip_list) != 0 p_base_ee_cur, q_base_ee_cur = self.controller.get_ee_pose() p_base_ee, q_base_ee, *_ = self.manip_list[0] diff_trans = np.linalg.norm(p_base_ee_cur - p_base_ee) diff_ori = 2 * np.arccos(min(abs(np.dot(q_base_ee_cur, q_base_ee)), 1.0)) pose_flag = np.logical_and( diff_trans < t_eps, diff_ori < o_eps, ) self.plan_flag = self.controller.num_last_cmd > 10 return np.logical_or(pose_flag, self.plan_flag) def is_done(self): if len(self.manip_list) == 0: return True if self.is_subtask_done(t_eps=self.skill_cfg.get("t_eps", 1e-3), o_eps=self.skill_cfg.get("o_eps", 5e-3)): self.manip_list.pop(0) return len(self.manip_list) == 0 def is_success(self): x, y, z = self.pick_obj.get_local_pose()[0] # pick_obj position within_boundary = ( self.place_boundary[0][0] <= x <= self.place_boundary[1][0] and self.place_boundary[0][1] <= y <= self.place_boundary[1][1] and self.place_boundary[0][2] <= z # <= self.place_boundary[1][2] ) print("pos :", self.pick_obj.get_local_pose()[0]) print("boundary :", self.place_boundary) print("within_boundary :", within_boundary) return within_boundary