A | B | C | D | E | F | G | H | I | J | K | L | M | N | O | P | Q | R | S | T | U | V | W | X | Y | Z | |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | Phases: | |||||||||||||||||||||||||
2 | intial | Starting pose of the robot arm at the very beginning of a grasp episode. | ||||||||||||||||||||||||
3 | approach | The first move in the grasp episode. Begins at the initial pose, and ends with the gripper in the workspace. This phase is separated from the other grasp phases, since the gripper is not visible at the beginning. During data collection, the target gripper pose for the approach phase was typically selected randomly. | ||||||||||||||||||||||||
4 | grasp | Main grasp phase consisting of multiple steps. Each step corresponds to recording an image, making a decision about the next pose, and beginning to execute a move to that pose. At the end of the last step, the gripper is closed. | ||||||||||||||||||||||||
5 | grasp/N | These sequence identifiers refer to steps in the grasp phase. | ||||||||||||||||||||||||
6 | gripper | Close the gripper at the end of the grasp episode. | ||||||||||||||||||||||||
7 | withdraw | Withdraw the closed gripper from the bin vertically upward, roughly 10 cm above the bin surface. | ||||||||||||||||||||||||
8 | move_to_present | After the withdraw phase, the robot moves the gripper in front of the camera to get a clean, unobstructed view of the object (this phase was skipped for some of the episodes and may not always be present). | ||||||||||||||||||||||||
9 | present | Presentation of grasped object to camera. | ||||||||||||||||||||||||
10 | raise | After presentation (or after withdraw, if present is skipped), the object is moved to a random position about 10 cm above the bin to drop the object. | ||||||||||||||||||||||||
11 | drop | Open the gripper and drop the object (if any object is held) into the bin. | ||||||||||||||||||||||||
12 | ||||||||||||||||||||||||||
13 | ||||||||||||||||||||||||||
14 | Feature Top Level Names: | Phase | The top level feature names describe an entity (e.g. robot or camera) or a phase or step of the action (e.g. grasp, approach). | |||||||||||||||||||||||
15 | approach | approach | ||||||||||||||||||||||||
16 | approach | approach | Pose reached at the end of the approach phase. | |||||||||||||||||||||||
17 | approach_sequence | approach | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
18 | camera | 4x4 transformation matrix from the base of the robot to the camera. This can be used if a camera calibration is needed. Note that camera calibration was not used in the closed-loop controller described in the paper, but is available for baselines. | ||||||||||||||||||||||||
19 | grasp | grasp | Commanded (target) pose for the last step in the grasp episode. | |||||||||||||||||||||||
20 | grasp | grasp | ||||||||||||||||||||||||
21 | grasp | grasp | Pose reached at the end of a step in the grasp episode. | |||||||||||||||||||||||
22 | grasp_sequence | grasp | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
23 | gripper | gripper | ||||||||||||||||||||||||
24 | gripper_sequence | gripper | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
25 | initial | initial | ||||||||||||||||||||||||
26 | move_to_present_sequence | move_to_present | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
27 | num_grasp_steps | grasp | Total number of steps (decisions) in the grasp episode, i.e. grasp/N has N in [0,...,num_grasp_steps - 1] | |||||||||||||||||||||||
28 | post_drop | drop | Images recorded after withdraw and raise but before the drop. | |||||||||||||||||||||||
29 | post_grasp | grasp | Images recorded right after the gripper closes. | |||||||||||||||||||||||
30 | pregrasp | grasp | Images recorded right before the gripper closes. | |||||||||||||||||||||||
31 | present | present | Images recorded during the presentation to the camera. | |||||||||||||||||||||||
32 | present_sequence | present | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
33 | raise_sequence | raise | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
34 | grasp | grasp | Reached (actual) pose for the last step in the grasp episode. May differ from the commanded pose due to imperfect end-effector control at high speed. | |||||||||||||||||||||||
35 | robot | Identifying information about the robot used to collect the data. Typically a robot ID number, which can be used to figure out which datapoints were collected by the same robots. | ||||||||||||||||||||||||
36 | withdraw_sequence | withdraw | Recorded robot status. Each sequence corresponds to robot telemetry (end effector poses, joint angles, recorded and commanded torques) recorded at about 20 Hz for the duration of that phase. | |||||||||||||||||||||||
37 | ||||||||||||||||||||||||||
38 | Feature Types: | The feature type describes the organization and interpretation of data within the feature. | ||||||||||||||||||||||||
39 | transforms/base_T_endeffector/vec_quat_7 | A pose is a 6 degree of freedom rigid transform represented with 7 values: vector (x, y, z) and quaternion (x, y, z, w). A pose is always annotated with the target and source frames of reference. For example, base_T_camera is a transform that takes a point in the camera frame of reference and transforms it to the base frame of reference. | ||||||||||||||||||||||||
40 | commanded_pose | Commanded pose is the input to the inverse kinematics computation, which is used to determine desired joint positions. | ||||||||||||||||||||||||
41 | reached_pose | Pose calculated by forward kinetics from current robot state in status updates. | ||||||||||||||||||||||||
42 | joint/commanded_torques | Commanded torques are the torque values calculated from the inverse kinematics of the commanded pose and the robot driver. | ||||||||||||||||||||||||
43 | joint/external_torques | External torques are torques at each joint that are a result of gravity and other external forces. These values are reported by the robot status query. | ||||||||||||||||||||||||
44 | joint/positions | Robot joint positions as reported by the robot status query. | ||||||||||||||||||||||||
45 | joint/velocities | Robot joint velocities as reported by the robot status query. | ||||||||||||||||||||||||
46 | depth_image | Depth image is encoded as an RGB PNG image where the RGB 24-bit value is an integer depth with scale 1/256 of a millimeter. | ||||||||||||||||||||||||
47 | image | Camera RGB images are stored in JPEG format. | ||||||||||||||||||||||||
48 | params | This is a simplified representation of a commanded robot pose and gripper status. These are the values that were solved for in the network. | ||||||||||||||||||||||||
49 | ||||||||||||||||||||||||||
50 | ||||||||||||||||||||||||||
51 | Feature Keys | |||||||||||||||||||||||||
52 | approach/params | |||||||||||||||||||||||||
53 | approach/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
54 | approach/reached_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
55 | approach_sequence/joint/commanded_torques | |||||||||||||||||||||||||
56 | approach_sequence/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
57 | approach_sequence/joint/external_torques | |||||||||||||||||||||||||
58 | approach_sequence/joint/positions | |||||||||||||||||||||||||
59 | approach_sequence/joint/velocities | |||||||||||||||||||||||||
60 | camera/transforms/camera_T_base/matrix44 | |||||||||||||||||||||||||
61 | grasp/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
62 | grasp/0/depth_image/encoded | |||||||||||||||||||||||||
63 | grasp/1/depth_image/encoded | |||||||||||||||||||||||||
64 | grasp/2/depth_image/encoded | |||||||||||||||||||||||||
65 | grasp/3/depth_image/encoded | |||||||||||||||||||||||||
66 | grasp/4/depth_image/encoded | |||||||||||||||||||||||||
67 | grasp/5/depth_image/encoded | |||||||||||||||||||||||||
68 | grasp/6/depth_image/encoded | |||||||||||||||||||||||||
69 | grasp/7/depth_image/encoded | |||||||||||||||||||||||||
70 | grasp/8/depth_image/encoded | |||||||||||||||||||||||||
71 | grasp/9/depth_image/encoded | |||||||||||||||||||||||||
72 | grasp/10/depth_image/encoded | |||||||||||||||||||||||||
73 | drop/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
74 | grasp/depth_image/encoded | |||||||||||||||||||||||||
75 | grasp/0/params | |||||||||||||||||||||||||
76 | grasp/1/params | |||||||||||||||||||||||||
77 | grasp/2/params | |||||||||||||||||||||||||
78 | grasp/3/params | |||||||||||||||||||||||||
79 | grasp/4/params | |||||||||||||||||||||||||
80 | grasp/5/params | |||||||||||||||||||||||||
81 | grasp/6/params | |||||||||||||||||||||||||
82 | grasp/7/params | |||||||||||||||||||||||||
83 | grasp/8/params | |||||||||||||||||||||||||
84 | grasp/9/params | |||||||||||||||||||||||||
85 | grasp/10/params | |||||||||||||||||||||||||
86 | grasp/0/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
87 | grasp/1/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
88 | grasp/2/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
89 | grasp/3/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
90 | grasp/4/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
91 | grasp/5/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
92 | grasp/6/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
93 | grasp/7/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
94 | grasp/8/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
95 | grasp/9/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
96 | grasp/10/commanded_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
97 | grasp/0/reached_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
98 | grasp/1/reached_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
99 | grasp/2/reached_pose/transforms/base_T_endeffector/vec_quat_7 | |||||||||||||||||||||||||
100 | grasp/3/reached_pose/transforms/base_T_endeffector/vec_quat_7 |