ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Moveit gets stuck after planning

asked 2017-07-02 10:06:19 -0600

Felix_N gravatar image

Hi everyone,

I currently want to send a joint-space goal to my 6DOF arm. move_point is a pointer to my move_group. The command comes from a GUI I wrote and the callback looks like this:

void goal_joint_callback(mybot::joint_command msg){
     std::vector<double> joint_group_positions(6);
     joint_group_positions[0] = msg.joint1;  // radians
     joint_group_positions[1] = msg.joint2;
     joint_group_positions[2] = msg.joint3; 
     joint_group_positions[3] = msg.joint4; 
     joint_group_positions[4] = msg.joint5;
     joint_group_positions[5] = msg.joint6;
     bool success = move_point->plan(*current_plan);
     ROS_INFO_NAMED("tutorial", "Visualizing plan 2 (joint space goal) %s", success ? "" : "FAILED");

The goal gets visualized quite nice in rviz and until then everything looks ok.

In my output terminal I can see "debug 1,2,3" but not "debug 4": I am also unable to send additional goals after the first one. I cant really tell why, because the same lines of code (bool success = move_point->plan(*current_plan);) do work at another place with the cartesian path.

Thank you in advance,


edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2017-07-03 03:08:49 -0600

v4hn gravatar image

Hey Felix_N,

my first guess is a problem with your control flow.

The plan function calls the ROS action provided by the move_group node and waits for the result. See here.

If you are already in a callback and don't have an additional ROS spinner, the action client will never receive the message it is waiting for.

Try to create (and start) a ros::AsyncSpinner in your main scope before ros::spin().

edit flag offensive delete link more


Probably related: #q264748 (see the comments).

gvdhoorn gravatar image gvdhoorn  ( 2017-07-03 03:10:52 -0600 )edit

You got it. I have a spinner everywhere else but forgot it at that point. Thanks.

Felix_N gravatar image Felix_N  ( 2017-07-05 15:35:52 -0600 )edit

Thanks, this was my issue as well, started and stopped an async spinner in my callback did the trick

amcelroy gravatar image amcelroy  ( 2017-07-27 15:28:23 -0600 )edit

@Felix_N and @amcelroy: please note what @v4hn writes:

create [..] a ros::AsyncSpinnerin your main scope

main scope != in your callback.

I would refrain from doing any spinning in callbacks. Only in very specific circumstances should this be done.

gvdhoorn gravatar image gvdhoorn  ( 2017-07-28 01:40:37 -0600 )edit

@Felix_N and @gvdhoorn - I do have an asyncspinner in my the main function that registers the callback, but for some reason I still had this exact same problem. My workflow in main is: - ros::init - allocate asyncspinner - start the spinner - register callback / do stuff - ros::waitForShutdown

amcelroy gravatar image amcelroy  ( 2017-07-28 20:22:05 -0600 )edit

Notice that waitForShutdown does not spin, so if your asyncspinner does not spin with more than one thread you will still see the problem.

v4hn gravatar image v4hn  ( 2017-07-29 03:47:21 -0600 )edit

Wow, ok, thanks so much. Creating AscynSpinner with 2 threads did the trick and allowed for the removal of the aysncspinner in the callback. Is it possible for myself or someone higher up to make a note of this on the Callbacks wiki?

amcelroy gravatar image amcelroy  ( 2017-07-29 08:56:50 -0600 )edit

Oh, I didn't know such a documentation actually exists :) I added a short comment to the section. Please have a look.

v4hn gravatar image v4hn  ( 2017-07-29 13:30:38 -0600 )edit

Question Tools



Asked: 2017-07-02 10:06:19 -0600

Seen: 1,120 times

Last updated: Jul 03 '17