oro_barrett_hw

Orocos Barrett Hardware Interface

Examples

For a 7-DOF WAM on bus 0 with a BHand

roslaunch test/dual_barrett.launch BUS:=0 7DOF:=true HAND:=true

For a 4-DOF WAM on bus 0 with a BHand and a 7-DOF WAM on bus 1:

roslaunch test/dual_barrett.launch LEFT_4DOF:=true LEFT_HAND:=true RIGHT_7DOF:=true

Zero-Calibrating the WAM Arm

Calibration of the WAM arm is done by storing the motor encoder angles at a given joint-space configuration. This involves preciscely positioning the arm manually while it is powered, but idle, and storing the encoder angles in a YAML file. Once these encoder angles have been stored, they can be used to determine the absolute position of each joint of the arm in a five-degree window around the desired joint position.

Using an existing calibration

The Orocos WAM component requires two properties to be set in order to set the position of the arm and switch from IDLE to RUN mode:

• home_position -- an exact joint-space configuration
• home_resolver_offset -- the motor-space encoder angles at the exact joint-space configuration

Given these two parameters, the arm can be precisely homed if it is manually positioned within five-degrees of the home_position and then initiailzed. Calling the wam.initialize service will use these parameters to determine the actual position of the wam.

The initialization operation is what is called by the Calibrate button in the GUI in the rqt_barrett package.

Saving a new calibration

The WAM robots should not need to be recalibrated frequently. They should only need to be recalibrated after re-tensioning, recabling, repairing, or if it appears that the calibration has drifted for some other reason.

In practice, the reading and saving of the encoder angles and can be done with the zerocal.py script.

1. Bring up the WAM robot you want to calibrate and make sure it is in IDLE mode.

2. Run the zerocal.py script in the namespace of the barrett manager:

   rosrun oro_barrett_hw zerocal.py path/to/config/for/wam_60.yaml __ns:=/barrett/barrett_manager
   

3. Manually position the arm to the exact home position using the appropriate jigs and levels.

4. Once the arm is exactly position, hit the enter key in the shell running zerocal.py and save the data to the appropriate yaml file.

Note that zerocal.py will overwrite the given file. While it will preserve YAML comments, it will re-order any block of parameters alphabetically.

Calibrating the WAM torque constants

Sometimes the WAM torque constants are incorrect. In order to compensate, the Orocos WAM component uses the torque_scales property. This is an array of scalar values which are used to multiply the output "raw" torque to the low-level driver.

The best way to calibrate these is by running the arm with a gravity compensation controller and making sure the arm can compensate for the weight of the links in various configurations.

It's also been observed that these torque constants can change slightly when using the arm for long periods of time. It may be dependent on the temperature of the motors, but normally it makes the motors weaker.