Arm Cartesian control

Reading the end-effector pose

Arm twins report the measured pose of their end-effector, so you can read where the tool actually is rather than inferring it from joint angles.

pose = twin.pose.get(version="v1")  # arms that publish /pose, e.g. AgileX Piper
print(pose.position.x, pose.position.y, pose.position.z)
print(pose.orientation.w)

The pose is in the robot’s base frame and updates continuously from the robot’s own joint feedback. Nothing is reported until the arm has sent its first feedback, so treat an empty result as “not ready yet” rather than “at the origin”. Where supported, the reported point is the grasp centre — the place the fingers actually close — not a point inside the wrist. Robots expose the offset to that point as a setting you can measure and correct on your own hardware.

Controlling speed

ee_move and ee_rotate_* accept an optional speed between 0 and 1, applied to that one command only. Use it to travel quickly and approach slowly.

twin.commands.ee_move(forward=0.10, speed=1.0)   # transit
twin.commands.ee_move(up=-0.02, speed=0.2)       # slow, careful descent

Omitting speed leaves the robot’s configured speed unchanged. A robot that exposes no speed control ignores the argument. grip, release, home, and base_rotate_* do not take a speed. On dual-arm robots the prefixed commands (left_ee_move, right_ee_rotate_up, …) accept it too.

Gripper width

Arms that report a continuous gripper opening send the measured width rather than a simple open/closed label, so you can tell whether a grasp actually holds an object by comparing the opening after closing against an empty close. See also Arm commands for the full command vocabulary.

Get started

Build with AI

Simulate & test

Connect hardware

Train & improve

Deploy & operate

Reading the end-effector pose

Controlling speed

Gripper width

​Reading the end-effector pose

​Controlling speed

​Gripper width

Reading the end-effector pose

Controlling speed

Gripper width