[update] update all robot configs to support all modes for both hands

dex_retargeting/configs/offline/ability_hand_left.yml

@@ -0,0 +1,16 @@
+retargeting:
+  type: position
+  urdf_path: ability_hand/ability_hand_left.urdf
+  wrist_link_name: "base_link"
+
+  target_joint_names: [ 'thumb_q1', 'thumb_q2', 'index_q1', 'middle_q1', 'pinky_q1', 'ring_q1' ]
+  target_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 20 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1
+
+  # To ignore the mimic joint tags in the URDF, set it to True
+  ignore_mimic_joint: False

dex_retargeting/configs/offline/allegro_hand_left.yml

@@ -0,0 +1,14 @@
+retargeting:
+  type: position
+  urdf_path: allegro_hand/allegro_hand_left.urdf
+  wrist_link_name: "wrist"
+
+  target_joint_names: null
+  target_link_names: [ "link_15.0_tip", "link_11.0_tip", "link_7.0_tip", "link_3.0_tip", "link_14.0",
+                       "link_10.0", "link_6.0", "link_2.0" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 2, 6, 10, 14 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1

dex_retargeting/configs/offline/inspire_hand_left.yml

@@ -0,0 +1,17 @@
+retargeting:
+  type: position
+  urdf_path: inspire_hand/inspire_hand_left.urdf
+  wrist_link_name: "base"
+
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  target_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 20 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1
+
+  # To ignore the mimic joint tags in the URDF, set it to True
+  ignore_mimic_joint: False

dex_retargeting/configs/offline/inspire_hand_right.yml

@@ -0,0 +1,17 @@
+retargeting:
+  type: position
+  urdf_path: inspire_hand/inspire_hand_right.urdf
+  wrist_link_name: "base"
+
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  target_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 20 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1
+
+  # To ignore the mimic joint tags in the URDF, set it to True
+  ignore_mimic_joint: False

dex_retargeting/configs/offline/leap_hand_left.yml

@@ -0,0 +1,13 @@
+retargeting:
+  type: position
+  urdf_path: leap_hand/leap_hand_left.urdf
+  wrist_link_name: "base"
+
+  target_joint_names: null
+  target_link_names: [ "thumb_tip_head", "index_tip_head", "middle_tip_head", "ring_tip_head", "thumb_dip", "dip", "dip_2", "dip_3" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 2, 6, 10, 14 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1

dex_retargeting/configs/offline/schunk_svh_hand_left.yml

@@ -0,0 +1,16 @@
+retargeting:
+  type: position
+  urdf_path: schunk_hand/schunk_svh_hand_left.urdf
+  wrist_link_name: "left_hand_base_link"
+
+  target_joint_names: [ 'left_hand_Thumb_Opposition', 'left_hand_Thumb_Flexion', 'left_hand_Index_Finger_Proximal',
+                        'left_hand_Index_Finger_Distal', 'left_hand_Finger_Spread', 'left_hand_Pinky',
+                        'left_hand_Ring_Finger', 'left_hand_Middle_Finger_Proximal', 'left_hand_Middle_Finger_Distal' ]
+  target_link_names: [ "left_hand_c", "left_hand_t", "left_hand_s", "left_hand_r",
+                        "left_hand_q", "left_hand_b", "left_hand_p", "left_hand_o", "left_hand_n", "left_hand_i"]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 20, 2, 6, 10, 14, 18 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1

dex_retargeting/configs/offline/shadow_hand_left.yml

@@ -0,0 +1,14 @@
+retargeting:
+  type: position
+  urdf_path: shadow_hand/shadow_hand_left.urdf
+  wrist_link_name: "ee_link"
+
+  target_joint_names: null
+  target_link_names: [ "thtip", "fftip", "mftip", "rftip", "lftip",
+                       "thmiddle", "ffmiddle", "mfmiddle", "rfmiddle", "lfmiddle" ]
+
+  target_link_human_indices: [ 4, 8, 12, 16, 20, 2, 6, 10, 14, 18 ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  # 1 means no filter while 0 means not moving
+  low_pass_alpha: 1

dex_retargeting/configs/teleop/inspire_hand_left.yml

@@ -0,0 +1,18 @@
+retargeting:
+  type: vector
+  urdf_path: inspire_hand/inspire_hand_left.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  target_origin_link_names: [ "base", "base", "base", "base", "base" ]
+  target_task_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+  scaling_factor: 1.15
+
+  # Source refers to the retargeting input, which usually corresponds to the human hand
+  # The joint indices of human hand joint which corresponds to each link in the target_link_names
+  target_link_human_indices: [ [ 0, 0, 0, 0, 0 ], [ 4, 8, 12, 16, 20 ] ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/inspire_hand_left_dexpilot.yml

@@ -0,0 +1,13 @@
+retargeting:
+  type: DexPilot
+  urdf_path: inspire_hand/inspire_hand_left.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  finger_tip_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+  scaling_factor: 1.15
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/inspire_hand_right.yml

@@ -0,0 +1,18 @@
+retargeting:
+  type: vector
+  urdf_path: inspire_hand/inspire_hand_right.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  target_origin_link_names: [ "base", "base", "base", "base", "base" ]
+  target_task_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+  scaling_factor: 1.15
+
+  # Source refers to the retargeting input, which usually corresponds to the human hand
+  # The joint indices of human hand joint which corresponds to each link in the target_link_names
+  target_link_human_indices: [ [ 0, 0, 0, 0, 0 ], [ 4, 8, 12, 16, 20 ] ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/inspire_hand_right_dexpilot.yml

@@ -0,0 +1,13 @@
+retargeting:
+  type: DexPilot
+  urdf_path: inspire_hand/inspire_hand_right.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: [ 'pinky_proximal_joint', 'ring_proximal_joint', 'middle_proximal_joint', 'index_proximal_joint',
+                        'thumb_proximal_pitch_joint', 'thumb_proximal_yaw_joint' ]
+  finger_tip_link_names: [ "thumb_tip",  "index_tip", "middle_tip", "ring_tip", "pinky_tip" ]
+  scaling_factor: 1.15
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/leap_hand_left.yml

@@ -0,0 +1,17 @@
+retargeting:
+  type: vector
+  urdf_path: leap_hand/leap_hand_left.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: null
+  target_origin_link_names: [ "base", "base", "base", "base" ]
+  target_task_link_names: [ "thumb_tip_head", "index_tip_head", "middle_tip_head", "ring_tip_head" ]
+  scaling_factor: 1.6
+
+  # Source refers to the retargeting input, which usually corresponds to the human hand
+  # The joint indices of human hand joint which corresponds to each link in the target_link_names
+  target_link_human_indices: [ [ 0, 0, 0, 0 ], [ 4, 8, 12, 16 ] ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/leap_hand_left_dexpilot.yml

@@ -0,0 +1,12 @@
+retargeting:
+  type: DexPilot
+  urdf_path: leap_hand/leap_hand_left.urdf
+  wrist_link_name: "base"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: null
+  finger_tip_link_names: [ "thumb_tip_head", "index_tip_head", "middle_tip_head", "ring_tip_head" ]
+  scaling_factor: 1.6
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/shadow_hand_left.yml

@@ -0,0 +1,17 @@
+retargeting:
+  type: vector
+  urdf_path: shadow_hand/shadow_hand_left.urdf
+  wrist_link_name: "ee_link"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: null
+  target_origin_link_names: [ "palm", "palm", "palm", "palm", "palm", "palm", "palm", "palm", "palm", "palm" ]
+  target_task_link_names: [ "thtip", "fftip", "mftip", "rftip", "lftip",  "thmiddle", "ffmiddle", "mfmiddle", "rfmiddle", "lfmiddle" ]
+  scaling_factor: 1.2
+
+  # Source refers to the retargeting input, which usually corresponds to the human hand
+  # The joint indices of human hand joint which corresponds to each link in the target_link_names
+  target_link_human_indices: [ [ 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 ], [ 4, 8, 12, 16, 20, 2, 6, 10, 14, 18 ] ]
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/configs/teleop/shadow_hand_left_dexpilot.yml

@@ -0,0 +1,12 @@
+retargeting:
+  type: DexPilot
+  urdf_path: shadow_hand/shadow_hand_left.urdf
+  wrist_link_name: "ee_link"
+
+  # Target refers to the retargeting target, which is the robot hand
+  target_joint_names: null
+  finger_tip_link_names: [ "thtip", "fftip", "mftip", "rftip", "lftip" ]
+  scaling_factor: 1.2
+
+  # A smaller alpha means stronger filtering, i.e. more smooth but also larger latency
+  low_pass_alpha: 0.2

dex_retargeting/constants.py

@@ -8,6 +8,7 @@ class RobotName(enum.Enum):
     svh = enum.auto()
     leap = enum.auto()
     ability = enum.auto()
+    inspire = enum.auto()
 
 
 class RetargetingType(enum.Enum):
@@ -27,6 +28,7 @@ class HandType(enum.Enum):
     RobotName.svh: "schunk_svh_hand",
     RobotName.leap: "leap_hand",
     RobotName.ability: "ability_hand",
+    RobotName.inspire: "inspire_hand",
 }
 
 ROBOT_NAMES = list(ROBOT_NAME_MAP.keys())

tests/test_optimizer.py

@@ -63,7 +63,7 @@ def generate_position_retargeting_data_gt(robot: RobotWrapper, optimizer: Positi
         return random_pin_qpos, init_qpos, random_target_pos
 
     @pytest.mark.parametrize("robot_name", ROBOT_NAMES)
-    @pytest.mark.parametrize("hand_type", [name for name in HandType][:1])
+    @pytest.mark.parametrize("hand_type", [name for name in HandType])
     def test_position_optimizer(self, robot_name, hand_type):
         config_path = get_default_config_path(robot_name, RetargetingType.position, hand_type)
 
@@ -105,7 +105,7 @@ def test_position_optimizer(self, robot_name, hand_type):
         assert np.mean(errors["pos"]) < 1e-2
 
     @pytest.mark.parametrize("robot_name", ROBOT_NAMES)
-    @pytest.mark.parametrize("hand_type", [name for name in HandType][:1])
+    @pytest.mark.parametrize("hand_type", [name for name in HandType])
     def test_vector_optimizer(self, robot_name, hand_type):
         config_path = get_default_config_path(robot_name, RetargetingType.vector, hand_type)
 
@@ -153,7 +153,7 @@ def test_vector_optimizer(self, robot_name, hand_type):
         assert np.mean(errors["pos"]) < 1e-2
 
     @pytest.mark.parametrize("robot_name", DEXPILOT_ROBOT_NAMES)
-    @pytest.mark.parametrize("hand_type", [name for name in HandType][:1])
+    @pytest.mark.parametrize("hand_type", [name for name in HandType])
     def test_dexpilot_optimizer(self, robot_name, hand_type):
         config_path = get_default_config_path(robot_name, RetargetingType.dexpilot, hand_type)