Ask Your Question

Oscillation when using orientation field of move_base with teb_local_planner.

asked 2018-10-29 12:45:52 -0600

mliefaard gravatar image

Hi everyone!

I am having an issue with the orientation fields of the move_base package and the teb_local_planner.

At first I would just set the orientation.w field to 1.0 and leave the xyz components set to 0. However, now I want to make the robot have a specific orientation when it reaches its goal location.

My train of thought was that if I wanted to rotate the robot around its base that I would have to rotate 0.5 * PI around its yaw axis. If I change the yaw component in my gazebo simulation then I can indeed rotate the robot around its y-axis.

I found this tutorial on the ROS wiki which states that you can convert a roll, pitch, yaw rotation to a quaternion.

Suppose my RPY respectively are [0, 0, 0.5 * PI] then the Quaternion is (w,x,y,z) (0.707107, 0, 0, 0.707107) move_base s canaccepts this as a valid quaternion and thus the goal is active. However, the robot starts to oscillate due to its local planner. I added an image of the RVIZ view where I visualized both the local and global plan. The robot stays on the first pose in the red poseArray and moves back and forward.

RVIZ view of the global and local plan.

Is my quaternion input incorrect or am I using the orientation field incorrectly? My goal is to have the robot arrive at its goal location with a specific orientation.

Here is the output of my navigation nodes:


 * /move_base/base_local_planner: teb_local_planner...
 * /move_base/global_costmap/footprint: [[0.25, 0.154], [...
 * /move_base/global_costmap/global_frame: map
 * /move_base/global_costmap/inflation_radius: 0.55
 * /move_base/global_costmap/obstacle_range: 2.5
 * /move_base/global_costmap/raytrace_range: 3.0
 * /move_base/global_costmap/robot_base_frame: base_link
 * /move_base/global_costmap/static_map: True
 * /move_base/global_costmap/update_frequency: 5.0
 * /move_base/local_costmap/footprint: [[0.25, 0.154], [...
 * /move_base/local_costmap/global_frame: map
 * /move_base/local_costmap/height: 10.0
 * /move_base/local_costmap/inflation_radius: 0.55
 * /move_base/local_costmap/obstacle_range: 2.5
 * /move_base/local_costmap/publish_frequency: 40.0
 * /move_base/local_costmap/raytrace_range: 3.0
 * /move_base/local_costmap/resolution: 1
 * /move_base/local_costmap/robot_base_frame: base_link
 * /move_base/local_costmap/rolling_window: False
 * /move_base/local_costmap/static_map: True
 * /move_base/local_costmap/update_frequency: 60.0
 * /move_base/local_costmap/width: 10.0
 * /robot_pose_ekf/debug: True
 * /robot_pose_ekf/freq: 50.0
 * /robot_pose_ekf/imu_used: True
 * /robot_pose_ekf/odom_used: True
 * /robot_pose_ekf/output_frame: map
 * /robot_pose_ekf/self_diagnose: False
 * /robot_pose_ekf/sensor_timeout: 1.0
 * /robot_pose_ekf/vo_used: False
 * /rosdistro: kinetic
 * /rosversion: 1.12.14

    link1_broadcaster (tf2_ros/static_transform_publisher)
    link2_broadcaster (tf2_ros/static_transform_publisher)
    map_server (map_server/map_server)
    move_base (move_base/move_base)
    robot_pose_ekf (robot_pose_ekf/robot_pose_ekf)


process[link1_broadcaster-1]: started with pid [19739]
process[link2_broadcaster-2]: started with pid [19740]
process[robot_pose_ekf-3]: started with pid [19758]
process[map_server-4]: started with pid [19776]
process[move_base-5]: started with pid [19841]
[ INFO] [1540834009.296069299, 13.907000000]: Loading from pre-hydro parameter style
[ INFO] [1540834009.310812145, 13.922000000]: Using plugin "static_layer"
[ INFO] [1540834009.317225170, 13.928000000]: Requesting the map...
[ INFO] [1540834009.520453286, 14.132000000]: Resizing costmap to 100 X 100 at 1.000000 m/pix
[ INFO] [1540834009.621709934, 14.232000000]: Received a 100 X 100 map at 1.000000 m/pix
[ INFO] [1540834009.626584742, 14.237000000]: Using plugin ...
edit retag flag offensive close merge delete


How are you giving the goals? The moving of the red poseArray is probably due to the planner re-planning the path. Teb local planner requires a lot of tuning and is not as intuitive as the other planners. You can try changing the goal tolerance parameters (yaw_goal_tolerance, xy_goal_tolerance)

pavel92 gravatar image pavel92  ( 2018-10-30 03:33:46 -0600 )edit

@pavel92 I am giving my goals using the actionlib interface for move_base:

typedef actionlib::SimpleActionClient<move_base_msgs::MoveBaseAction> MoveBaseClient;

This is the correct way to give movement goals right?

I am currently looking into changing the parameters, thanks for your response

mliefaard gravatar image mliefaard  ( 2018-10-30 05:41:47 -0600 )edit

You can give navigation goals directly in RViz using the 2D Nav Goal and allows you to easily set the desired orientation

pavel92 gravatar image pavel92  ( 2018-10-30 06:58:52 -0600 )edit

Yes, the navigation goals set directly in the RViz interface do not cause the same issue. I did not know that that was using a 2D Nav Goal so I am going to try to use 2D Nav Goal from my C++ node and see if I encounter the same issue as with the actionlib interface.

mliefaard gravatar image mliefaard  ( 2018-10-30 08:40:16 -0600 )edit

The issue still persists if I use 2D Nav Goals published to move_base_simple/goal. Sometimes the teb_local_planner just does not generate a valid PoseArray to follow. It seems like it is trying to strafe as seen in the image that I added. Could this be caused by my parameters for teb_local_planner?

mliefaard gravatar image mliefaard  ( 2018-10-30 09:51:15 -0600 )edit

can you please edit your question and add the teb_local_planner params?

croesmann gravatar image croesmann  ( 2018-10-30 10:36:00 -0600 )edit

If the robot rotates to the desired orientation with the goals set in RViz that means that probably the problem is not coming from the goal tolerance parameters from teb_local_planner params but may be coming from different parameters so please post your teb_local_planner_params.yaml file

pavel92 gravatar image pavel92  ( 2018-10-31 03:36:27 -0600 )edit

Sorry for the late response. I spent a lot of time configuring teb_local_planner but the problem persisted so I eventually decided to try other planners as pavel92 suggested earlier. eband_local_planner is working perfectly for me. Thanks for pointing me in the right direction!

mliefaard gravatar image mliefaard  ( 2018-10-31 14:02:13 -0600 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2018-10-31 14:04:53 -0600

mliefaard gravatar image

updated 2018-10-31 14:05:23 -0600

After a lot of debugging it turned out that my teb_local_planner configuration was the problem since the problem also persisted when I issued a goal in RViz instead of from code.

As @pavel92 noted, teb_local_planner requires a lot of fine tuning. So I eventually gave up and switched to eband_local_planner which works perfectly for my use case. In my experience eband_local_planner was way simpler to configure and it worked immediately for my diff_drive robot.

(Unfortunately I cannot mark this as answered because I don't have enough points)

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2018-10-29 12:42:30 -0600

Seen: 549 times

Last updated: Oct 31 '18