Revision history [back]

JointState msgs in ROS are typically only used for reporting the state of joints, not for commanding them / sending new setpoints (I know of some exceptions, but am not sure whether I would call that a 'recommended' practice).

For executing trajectories, you'll want to look into setting up a FollowJointTrajectory action server-client pair. See control_msgs for the action definition (specifically: FollowJointTrajectory). See joint_trajectory_action for an old / the PR2 implementation (you probably don't want to use that node, but it does help to understand JointTrajectory action server interaction).

For newer setups, I would perhaps look at ros_control, and then specifically at the joint_trajectory_controller. That exposes the required interfaces, and also allows you to work with Gazebo simulated robots and real hardware (almost) seamlessly.

JointState msgs in ROS are typically only used for reporting the state of joints, not for commanding them / sending new setpoints (I know of some exceptions, but am not sure whether I would call that a 'recommended' practice).

For executing trajectories, you'll want to look into setting up a FollowJointTrajectory action server-client pair. See control_msgs for the action definition (specifically: FollowJointTrajectory). See joint_trajectory_action for an old / the PR2 implementation (you probably don't want to use that node, but it does help to understand JointTrajectory action server interaction).

For newer setups, I would perhaps look at ros_control, and then specifically at the joint_trajectory_controller. That exposes the required interfaces, and also allows you to work with Gazebo simulated robots and real hardware (almost) seamlessly.seamlessly (through gazebo_ros_control).