**Input:** - sceneIds: int or list of int of the scene ids. **Output:** - a list of int of the object ids in the g

**Output:** - center, open_point, upper_point, each of them is a numpy array of shape (2,)

Gets the transformation that puts the object resting exactly on the z=delta plane. Parameters ---------- T_obj_surfac

**Input:** - centers: np.array(-1,3) of the translation - Rs: np.array(-1,3,3) of the rotation matrix - widths: np.array(-1) of th

- return the config dict

**Output:** - center, open_point, upper_point, each of them is a numpy array of shape (2,)

Calls get_mat() to get the 4x4 matrix.

**Output:** - Convert self.x, self.y, self.z, self.alpha, self.beta and self.gamma into mat_4x4 pose

**Input:** - scene_id: int of the scen index. - ann_id: int of the annotation index. **Output:** - obj_li

**Output:** - Convert self.alpha, self.beta, self.gamma into self.quat

return models in model coordinate.

**Output:** - Convert self.x, self.y, self.z into self.translation

**Input:** - centers_xyz: numpy array of the center points of shape (-1, 3). - open_points_xyz: numpy array of the open points of shape

**Input:** - center_xyz: numpy array of the center point. - open_point_xyz: numpy array of the open point. - upper_point_xyz: nump

**Input:** - sceneId: int of the scene index. - camera: string of type of camera, 'realsense' or 'kinect' - annId:

**Input:** - sceneId: int of the scene index. - camera: string of type of camera, 'realsense' or 'kinect' - annId:

**Input:** - sceneId: int of the scene index. - camera: string of type of camera, 'realsense' or 'kinect' - annId:

**Input:** - sceneId: int of the scene index. - camera: string of type of camera, 'realsense' or 'kinect' - annId:

**Input:** - data_path: path to load .obj & .sdf files **Output:** - obj: dexnet model

Author: chenxi-wang **Input:** - numpy array of shape (3, 3) of the rotation matrix. **Output:** - binormal: numpy array of s

Computes the moment arm to a point x. Parameters ---------- x : 3x1 :obj:`numpy.ndarray` point to get moment arm

**Input:** - translation_thresh: float of the translation threshold. - rotation_thresh: float of the rotation threshold.

Normalize the mesh's orientation along its principal axes. Transforms the vertices and normals of the mesh such that the origin of th

Regression loss for offsets (z, w, theta) in Multi-Grasp Generator. :param regs: offset [B, anchor_k, H/r, W/r] :param gt_regs: ground truth

**Input:** - key_points: np.array(-1,4,3) of grasp key points, definition is shown in key_points.png - scores: numpy array of batch gra

pose = pos_quat_to_pose_4x4(pos, quat) Convert pos and quat into pose, 4x4 format Args: pos: length-3 position quat: length-4 quaternion

Convert pose, 4x4 format into pos and quat. Args: pose: numpy array, 4x4 Returns: pos: length-3 position quat: le

**Input:** - pose_vector: A list in the format of [id,x,y,z,alpha,beta,gamma] **Output:** - A pose class instance

**Input:** - Pose vector list defined in xmlhandler.py **Output:** - list of poses.

Returns the maximal span of the mesh's coordinates. The maximal span is the maximum coordinate value minus the minimal coordinate val

Compute grasp quality. Parameters ---------- grasp : :obj:`GraspableObject3D` grasp to quality quality on

Reads in the SDF file and returns a Sdf object. Returns ------- :obj:`Sdf` A Sdf created from the data in the fil

**Input:** - index: list of the index of grasp

Remove invisible part of object model according to scene point cloud. Input: cloud: [np.ndarray, (N,3), np.float32] scene poi

Remove any vertices that are not part of a triangular face. Note ---- This method will fail if any bad triangles are present,

Returns the sdf surface normal at the given coordinates by computing the tangent plane using SDF interpolation. Parameters --

Projects the local surface onto the tangent plane at a contact point. Parameters ---------- width : float

**Input:** - camera: string of type of camera, 'kinect' or 'realsense'. - depths: numpy array of the depths image. - depth_method:

**Input:** - color: optional, tuple of shape (3) denotes (r, g, b), e.g., (1,0,0) for red **Ouput:** - list of ope

**Output:** - list of open3d.geometry.Geometry of the grippers.

**input:** - opencv_rgb: numpy array of opencv BGR format. - numGrasp: int of the number of grasp, 0 for all. **Output:** - n

**input:** - opencv_rgb: numpy array of opencv BGR format. **Output:** - numpy array of opencv RGB format that sho

**Input:** - camera: string of type of camera, 'realsense' or 'kinect'. **Output:** - RectGraspGroup instance or None.

Get the torques that can be applied by a set of force vectors at the contact point. Parameters ---------- forces : 3x

Return the total volume of the mesh. Returns ------- float The total volume of the mesh.

train one epoch. Args: epoch (int): epoch idx anchornet (nn.Module): anchornet (GHM) localnet (nn.Module): localnet (NMG)

Returns a new SDF transformed by similarity tf.

Returns an array of the triangle centers as 3D points. Returns ------- :obj:`numpy.ndarray` of :obj:`numpy.ndarray` of float

Returns a list of the triangle normals. Parameters ---------- align_to_hull : bool If true, we re-orient the norm

Author: chenxi-wang **Input:** - dataset_root: str, graspnet dataset root - scene_name: str, name of scene folder, e.g. scene_0000

**Input:** - points: (N, 3) **Output:** - points: (n, 3)

Width of grasp from endpoints.

Writes a Mesh3D object out to a .obj file format. Parameters ---------- mesh : :obj:`Mesh3D` The Mesh3D object to

Rigid transformation from grasp frame to object frame. Rotation matrix is X-axis along approach direction, Y axis pointing between the

Return pose.

Makes a copy of the config.

Makes a copy of the config.

Check equivalence by rotation about the z axis.