xarm_driver publishes joint_states even though joint state controller should instead

[BryanStuurman]

xarm_api/src/xarm_driver.cpp creates a publisher for the topic "joint_states" on line 236: https://github.com/xArm-Developer/xarm_ros/blob/master/xarm_api/src/xarm_driver.cpp#L236 However this functionality is preferentially provided by the ros joint_state_controller, and solutions that use that (I note xarm_ros no longer starts it) end up with colliding messages, which resulted in awful velocity data due to some sample rate mismatches ( some of that can be seen here: https://github.com/xArm-Developer/xarm_ros/issues/96)

Additionally I've recently updated to firmware 2.01 on one robot and i see that the velocity data from xarm_hw.read() is now discretized at 10Hz instead of 100Hz regardless of the read back rate. Can the old faster rate be had somehow? seems like i spoke too soon, I can see smooth data.

[BryanStuurman]

Here's an example of the difference in functionality when the joint_state_controller is reactivated, patched (https://github.com/ros-controls/ros_controllers/commit/d14d9023e899d7103953d275e207829e60fec8e4) and the built in joint_states publisher is moved to "joint_states2":

Trajectory velocity

Slow manual mode motion

Fast jerks in manual mode

[penglongxiang]

Hi @BryanStuurman, thanks for the feedback. To make it clear, you are suggesting on re-activating the joint_state_controller to publish "joint_states" topic and remove the same topic publisher in xarm_driver node because of its low resolution (5Hz by default setting) and delay, is that right? We will take a look into this issue and reply to you later.

[BryanStuurman]

Yes, that is my suggestion. Low resolution and delay are contributing factors, and because this driver is for use with ros_control, and the joint_state_controller is the expected interface that is used for it. Like other controllers it means that all of the expected features are available, like proper namespaces, management with the controller manager interface, and the ability to load custom variants.

[penglongxiang]

Hi @BryanStuurman we have made the changes as you suggested, please update the repository and see if it behaves as expected.

[BryanStuurman]

This is looking good.