Navigation stack never reach goal

asked 2011-08-27 04:15:24 -0600

429 ●8 ●12 ●22

updated 2013-09-09 04:15:09 -0600

Hi everybody !

I have some trouble implementing the navigation stack for my own custom robot. It has an arduino which do the position estimation and send it to ROS inside my computer. The TF and odom messages are OK, I display them in RVIZ and the position is fine. Then I do a "dummy laser" which always send a laser a 3meters (I don't want to use a laser for now).

=> My robot in RVIZ turning on himself.

The global planner seems to be OK, the path to go the goal is fine, but the robot does not follow the line, so mayby the problem come from the local planner.

Anyone ever had this problem ? Or do you think what can be wrong in my configuration ?

Here is my setup for the navigation stack :

base local planner :

TrajectoryPlannerROS:

  max_vel_x: 0.1
  min_vel_x: 0.01
  max_rotational_vel: 1.0
  min_in_place_rotational_vel: 0.4

  acc_lim_th: 0.05
  acc_lim_x: 0.005
  acc_lim_y: 0.0

  holonomic_robot: false

costmap common param :

obstacle_range: 2.5
raytrace_range: 3.0
robot_radius: 0.1
inflation_radius: 0.4

observation_sources: laser_scan_sensor

laser_scan_sensor: {sensor_frame: laser_frame, data_type: LaserScan, topic: /scan, marking: true, clearing: true}

global costmap :

global_costmap:
  global_frame: map
  robot_base_frame: base_link
  update_frequency: 3.0
  static_map: false
  rolling_window: true
  width: 3.0
  height: 3.0
  resolution: 0.005
  origin_x: -1.5
  origin_y: -1.5

local costmap :

local_costmap:
  global_frame: odom
  robot_base_frame: base_link
  update_frequency: 4.0
  publish_frequency: 4.0
  static_map: false
  rolling_window: true
  width: 3.0
  height: 3.0
  resolution: 0.01
  origin_x: -1.5
  origin_y: -1.5

odom and map are the same thing.

I have also sometime this warning : "Control loop missed its desired rate of 20.0000Hz... the loop actually took 0.2422 seconds"

Thanks,

Jbot

edit retag flag offensive close merge delete

Comments

Instead of a dummy laser, it would be better to not use a laser at all. The data you are simulating is not at all realistic (unless your robot is encased in a 3meter radius bubble that moves with the robot).

Eric Perko ( 2011-08-27 13:16:31 -0600 )edit

Can you update your rviz video to include a visualization of the obstacles and such? See http://www.ros.org/wiki/navigation/Tutorials/Using%20rviz%20with%20the%20Navigation%20Stack for details on recommended rviz setups for navigation.

Eric Perko ( 2011-08-27 13:18:46 -0600 )edit

Not an answer to the question, but if you are getting lots of messages about the control loop missing, it may be wise to reduce your control loop speed to about 10 Hz, or possibly decrease the sizes of your costmap. We have found that missing the control loop rate can cause some problems.

mjcarroll ( 2011-08-28 03:20:22 -0600 )edit

Ok, thanks. Putting the control loop speed to 10Hz is better. But the robot has "un-smooth" movements. Maybe the tf from the arduino is not refreshed often enought.

Jbot ( 2011-09-05 02:57:08 -0600 )edit

add a comment

answered 2011-12-08 05:23:31 -0600

This post is a wiki. Anyone with karma >75 is welcome to improve it.

Ok, thanks. Putting the control loop speed to 10Hz is better. But the robot has "un-smooth" movements. Maybe the tf from the arduino is not refreshed often enought. - Jbot (Sep 05)

edit flag offensive delete link

Comments

Setting the control loop speed is controlled with controller_frequency private param of the move_base node: (angle bracket) param name="controller_frequency" value="10.0"