Pose array's variance converge(navigation stack, amcl)

February 4, 2013, 12:53 pm

≫ Next: The size of particlecloud in PoseArray Message

≪ Previous: amcl not publishing map -> odom

$

0

0

Hi, I'm using nav stack with my robot.
I'd like to set a final variance of a robot pose array.
I set amcl's parameters "update_min_a" and "update_min_d" to 0.0
because I'd like to have the robot estimate its pose
even if it stops.
In this case, a pose array showen on rviz converges as soon as amcl node runs.
It's what I like, but the array over-converges.
I mean, all arrows showing robot's pose lie on top of each other.
I guess that some threshold is offered to avoid that.
However, I changed all parameters listed in amcl's wiki and couldn't find
which one is it.
Could you tell me which parameter is used to avoid the above problem?
Thanks in advance.

---

The size of particlecloud in PoseArray Message

September 27, 2014, 12:22 pm

≫ Next: use navigation stack with different method for localization

≪ Previous: Pose array's variance converge(navigation stack, amcl)

$

0

0

Hi, I wrote a node that subscribes to amcl particlecloud. I would like to write a callback function that process all the Pose array elements. How can i get the size of the particlecloud or specifically the number of poses in the PoseArray Message?

use navigation stack with different method for localization

October 2, 2014, 9:19 am

≫ Next: Can't navigate and use camera at the same time

≪ Previous: The size of particlecloud in PoseArray Message

$

0

0

Hello everyone, I am pretty sure that it is possible to use the navigation stack ([http://wiki.ros.org/navigation](http://wiki.ros.org/navigation)) with another localization method than AMCL ([http://wiki.ros.org/amcl](http://wiki.ros.org/amcl)) since I have seen that people already tried to use hector SLAM. However, I was not able to understand how they realized this "switch". Does anyone have an idea where I should start? I have been using the navigation stack with some success but wanted to try out another localization technique. Cheers Marcus

Can't navigate and use camera at the same time

October 24, 2014, 1:03 am

≫ Next: navigation with husky on Rviz and Gazebo : robot sliding and odom problem

≪ Previous: use navigation stack with different method for localization

$

0

0

I am working on a project that requires both navigation and object recognition modules. The problem I have is that modules work when they are run individually but as soon as I try to run them simultaneously then one fails. I I start amcl first I can do the navigation but there is no data from the /camera/rgb/image_mono (although there is data on the /camera/ir/image_raw topic) topic even after running roslaunch openni_launch openni.launch. If I start openni.launch first then in amcl I get the error : extrapolation error looking up robot pose. Can I not do both at the same time when using the /camera/rgb/image_mono topic? I am using the Kinect Sensor that comes with the turtlebot for both recognition and navigation. The ros distribution is groovy.

navigation with husky on Rviz and Gazebo : robot sliding and odom problem

October 21, 2014, 7:02 am

≫ Next: Using PointCloud in AMCL

≪ Previous: Can't navigate and use camera at the same time

$

0

0

Hello, I am trying to navigate in a simple closed world with a husky robot, using move_base. I mapped the world with hector_mapping. When I navigate with move_base, the /encoder topic published by the ekf doesn't match with the /odom topic (/odom_combined). Here are the frames : map -> odom_combined -> base_footprint -> base_link - map -> odom_combined : own broadcaster ("manual") - odom_combined -> base_footprint : robot_pose_ekf - base_footprint -> base_link : robot_state_publisher AND map-> odom : amcl Launching move_base: (and parameters) Here are the notable arguments of localization (amcl) launching file : The other point is that the robot doesn't go straight away but always slides when publishing : $rostopic pub -r 4 /husky/cmd_vel_unsafe geometry_msgs/Twist '{linear: {x: 1.0, y: 0.0, z: 0.0 }, angular: {x: 0.0, y: 0.0, z: 0.0 } }' Any idea? Is it from gazebo world or from parameters of the robot? Thank you

Using PointCloud in AMCL

October 27, 2014, 5:32 am

≫ Next: navigation stack get crash

≪ Previous: navigation with husky on Rviz and Gazebo : robot sliding and odom problem

$

0

0

Hi, I am using the AMCL filter from ROS packages and I have configured it following the instructions on the website. I am wondering if it is possible to change the sensor input from the laser to a pointcloud without using any other package. Thank you

navigation stack get crash

October 14, 2014, 6:29 am

≫ Next: How to localize a robot in a known map

≪ Previous: Using PointCloud in AMCL

$

0

0

Hi, I am working with the navigation stack and when I command a goal (by rviz) which is inside an obstacle I get the following error. This is something I have begun to experience recently (after upgrading some Ros packages). If you need any configuration file, I can provide them. [ WARN] [1413287442.128433137, 500.284000000]: Map update loop missed its desired rate of 5.0000Hz... the loop actually took 1.0820 seconds [ WARN] [1413287447.391084978, 502.616000000]: Map update loop missed its desired rate of 5.0000Hz... the loop actually took 2.1320 seconds [ WARN] [1413287450.041637304, 503.840000000]: Map update loop missed its desired rate of 5.0000Hz... the loop actually took 1.2240 seconds [ WARN] [1413287452.804671449, 505.062000000]: Map update loop missed its desired rate of 5.0000Hz... the loop actually took 1.0220 seconds [ WARN] [1413287452.835355243, 505.075000000]: Clearing costmap to unstuck robot. [move_base-3] process has died [pid 16397, exit code -11, cmd /home/summitxl/ros_catkin_ws/install_isolated/lib/move_base/move_base odom:=odometry/filtered hokuyo_laser_topic:=scan cmd_vel:=/summit_xl_robot_control/command __name:=move_base __log:=/home/summitxl/.ros/log/3bb2ad12-5396-11e4-ac48-9cb70d1c3f04/move_base-3.log]. log file: /home/summitxl/.ros/log/3bb2ad12-5396-11e4-ac48-9cb70d1c3f04/move_base-3*.log Thank you My debug output is here: https://gist.github.com/arenillas/2408efc78094d1a0be24 EDIT: Using the parameter default_tolerance (5 meters) the result is: [ WARN] [1414505173.888202638, 61.526000000]: Map update loop missed its desired rate of 5.0000Hz... the loop actually took 0.7650 seconds [ WARN] [1414505173.930135630, 61.537000000]: Invalid Trajectory 0.000000, 0.000000, -0.400000, cost: -1.000000 And after some time and repetitions of these two warnings: [ WARN] [1414505212.929234900, 71.937000000]: Clearing costmap to unstuck robot. [move_base-3] process has died [pid 3922, exit code -11, cmd /home/summitxl/ros_catkin_ws/install_isolated/lib/move_base/move_base odom:=odometry/filtered hokuyo_laser_topic:=scan cmd_vel:=/summit_xl_robot_control/command __name:=move_base __log:=/home/summitxl/.ros/log/24466806-5eab-11e4-bdf7-047d7b44770f/move_base-3.log]. log file: /home/summitxl/.ros/log/24466806-5eab-11e4-bdf7-047d7b44770f/move_base-3*.log EDIT: Finally, it was a problem in costmap package. It has been fixed.

How to localize a robot in a known map

October 2, 2014, 10:05 pm

≫ Next: Getting the navigation goal from RVIZ

≪ Previous: navigation stack get crash

$

0

0

Hi everyone, I'm new in this world of mobile robots and ROS. Currently I wanna know if there are techniques implemented in ROS for self-localization or localization of a robot in a known map that it's provided by a map_server. I try to use AMCL from de Navigation Stack but I could not make it work because I use hector_mapping without odometry information to build a accurate map and, in this moment, my robot doesn't provide me the odometry information. I just wanna know if anyone in this community has/meets or has developed a technique ir ros to do this requirement for navigating. I configured my amcl.launch like several examples in this community but AMCL doesn't work. I tried this solution (http://answers.ros.org/question/104971/using-hector-slam-with-amcl-for-localization-in-a-pre-made-map/) and unfortunately I could not get good results, in RVIZ the particleCloud topic doesn't provide a real good localization, any error or warning doesn't appear in the console and all particles, provide in this topic, diverge against to converge to a good pose. I don't know what to do. I think I have to provide a good odometry from my robot but I have read that if I provide bad odometry, AMCL won't work. Anyone knows or has a good tutorial to do that, the information from AMCL package is still insufficient for me because I'm a noob in these topics. Please, help me, I would appreciate any idea, comment or any solution to do this.

---

Getting the navigation goal from RVIZ

November 3, 2014, 9:06 am

≫ Next: amcl localisation on custom made bot

≪ Previous: How to localize a robot in a known map

$

0

0

I am able to get the initial pose of my turtlebot using : `rospy.wait_for_message('initialpose', PoseWithCovarianceStamped)` `rospy.Subscriber('initialpose', PoseWithCovarianceStamped)` This allows me to get a the initial pose when a user clicks sets the 2d Pose estimate in RVIZ. My question is threefold - How do I achieve the same with the 2d Navigation Goal (getting it fom RVIZ)? - What message should I wait for and subscribe to? - What TYPE of message am I looking for? For example with initialpose I get a PoseWithCovarianceStamped message.

amcl localisation on custom made bot

November 4, 2014, 1:56 pm

≫ Next: Can turtlebot_simulator be loaded with a PGM map?

≪ Previous: Getting the navigation goal from RVIZ

$

0

0

Hi! This is part of my final year project work. I am facing multiple problems and i need help. I am aiming to build a robot that can localize itself on a known map and navigate to certain given points on the map. The robot has the basic structure as mentioned in navigation [stack tutorial](http://wiki.ros.org/navigation/Tutorials/RobotSetup/TF). Here for publishing the odometry, I have used wheel encoders and receiving them from an arduino board via serial. I am publishing a already generated map using `map_server` . In order to localize the robot on this map i am planning to use `amcl`. On rviz, when i set the fixed frame as `/map` it displays the correct map but shows an error that fixed frame doesnot exist. For publishing the transform i have used exactly the same code as mentioned on [tf_broadcaster tutorial](http://wiki.ros.org/navigation/Tutorials/RobotSetup/TF). How to get amcl working so that i can localise my robot?? I am getting laser scans on topic `/scan`. - Is it necessary to have a urdf model of robot before i can do localisation? - How can i transform data from this /scan topic to base_link as mentioned in tutorial?

Can turtlebot_simulator be loaded with a PGM map?

November 17, 2014, 10:56 am

≫ Next: repeating the simulation

≪ Previous: amcl localisation on custom made bot

$

0

0

Hello, I have a Turtlebot 2 on ROS Hydro, and I am working on an algorithm for smart autonomous navigation through a known map. I have a map of the building I wanted to test in PGM form, obtained from running Gmapping. In real life, I would take this map, boot up my 'bot, load AMCL for localization within my PGM map, and then start my navigation algorithm. I'd like to also be able to do this in the simulation world, i.e., start the simulator, load my map into Gazebo, start AMCL with the same map, and start my algorithm and watch the simulated robot drive around. Currently I cannot find any way to load the PGM into Gazebo or convert it into a .world file. Is this something that's already available or would I need to write a custom program to create a world file where each black pixel is an obstacle? Thanks in advance!

repeating the simulation

May 5, 2014, 7:48 pm

≫ Next: Which is each frame?

≪ Previous: Can turtlebot_simulator be loaded with a PGM map?

$

0

0

Hi, I made a localization simulation in ROS using stage and amcl running from launch file. I would like to repeat the simulation for 1000 times, is there an easy way, tools or tricks to do that? BR Anas

Which is each frame?

December 16, 2014, 8:40 am

≫ Next: Exclude topic from namespace

≪ Previous: repeating the simulation

$

0

0

I have a frame tree in which I have /odom and /map frames. My EKF publishes in /odom frame. My AMCL algorithm (I think) creates a transformation between /odom and /map frames. Gazebo gives me the modelStates positions. The problem is that it doesn't specify a frame. By default, it is set to be world_frame. I want to check how good is my AMCL. Because of that I think I should transform my /odom position given by EKF to /map frame and there compare it to gazebo's measurements. However, I don't know where Gazebo is referencing the position. [QUESTION UPDATE] I have two issues that are explined below: 1) Without any AMCL (and without having /map frame in the tree) I used the Gazebo ModelStates topic to check the EKF result assuming that both were published in /odom frame. Was it Ok? 2) Now I have an AMCL running and because of it /map frame appears in the tree. I want to check how good is my AMCL algorithm comparing the pose shown in /amcl_pose (which is in /map frame) with /gazebo/ModelStates topic. How can I do that? I don't know which frame does gazebo use to reference the data in topic ModelStates. Thank you very much

Exclude topic from namespace

January 21, 2015, 6:00 am

≫ Next: Problems in path planning and amcl

≪ Previous: Which is each frame?

$

0

0

Hello, I have a Turtlebot running the navigation stack and a workstation that runs the corresponding map-server. The amcl node is launched in a namespace so that all published and subscribed topics get prefixed with that namespace. So amcl is subscribing to /namespace/map instead of just /map. Now I want to exclude the map topic from the namespace so that amcl subscribes to /map again but all other topics should still be prefixed. I tried to remap it from "map" to "/map" but that didnt't change anything. Thanks for your help.

Problems in path planning and amcl

January 21, 2015, 9:55 am

≫ Next: How to predict the behavior of Autonomous Navigator of ROS?

≪ Previous: Exclude topic from namespace

$

0

0

I am using gazebo to simulate a SummitXL. The mission is to explore an area. To do that I have developed an algorithm that generates points equally distributed around the area. I use these points as goals for path planning algorithm. Ii is not working properly: First of all, in the AMCL, it obtains a great estimation but when the robot continues its movement, the AMCL gets a worse estimation because all the particles are too close. The most important issue is explained here: When the robot is in a corridor it cannot go through it because it thinks it is closed. I think that this is happening due to the amcl. The robot has the local map with an obstacle (dark grey in the picture) and the global map with a false obstacle (light grey in the picture)  The global map is wrong and it should be une meter away, to coincide with the local map. If you can help me I would appreciate it. I can give more data if you want. I will be updating the question as long as I know more important facts about what is happening.

---

How to predict the behavior of Autonomous Navigator of ROS?

January 20, 2015, 11:59 pm

≫ Next: How to navigate without localization

≪ Previous: Problems in path planning and amcl

$

0

0

Hi, 1. I am specifically interested in predicting the next move of autonomous navigator of ROS. 2. What should be the packages, I need to look at. I have gone through [this page](http://wiki.ros.org/navigation/Tutorials/RobotSetup) ? 3. I f I want to modify the behavior of autonomous navigator of ROS; in which direction I should proceed. **Many thanks** for any kind of help.

How to navigate without localization

January 26, 2015, 11:11 pm

≫ Next: amcl not reacting to /initialpose topic

≪ Previous: How to predict the behavior of Autonomous Navigator of ROS?

$

0

0

We're using **turtlebot_gazebo** now and every time we **navigate**, it uses **amcl**. However, in our task,robot just needs to **avoid the obstacles and follow the target person in a small range of view**. In short, we want to modify or some way to **navigate without localization**. I suspect it is sth related to the tf and move_base but not quite clear what to do. Can it be easily fulfilled? What should we do? *Many Thanks!*

amcl not reacting to /initialpose topic

February 5, 2015, 3:44 pm

≫ Next: turtlebot position fluctuates

≪ Previous: How to navigate without localization

$

0

0

Hi folks, I've set up navigation stack for my robot and all the frames seem to be correct (according to my tf_three, and frames...map->odom->base_link, etc). if I move the robot, I can see it moving in the map as well as the updated feed from all sensors. The problem I am having is that AMCL is subscribed to the /initialpose topic, but when I give the robot an initial pose (using the "2D Pose Estimate" button), the robot does not move in rviz. When I do "rostopic echo /initialpose" I can see the data changing, but the robot image does not move in rviz.

turtlebot position fluctuates

February 25, 2015, 3:53 pm

≫ Next: Running robot_pose_ekf and amcl

≪ Previous: amcl not reacting to /initialpose topic

$

0

0

Dear All, Please have look on this video: http://youtu.be/MiTcy0rHY78 I'm running turtlebot gazebo simulator, but as you see the pose of robot has lots of error and fluctuates I have run the following: 1)roslaunch turtlebot_gazebo turtlebot_world.launch 2)roslaunch turtlebot_gazebo gmapping_demo.launch 3)roslaunch turtlebot_rviz_launchers view_navigation.launch 4)roslaunch turtlebot_gazebo amcl_demo.launch The position of the robot in gazebo is fixed, but the localized position in rviz is fluctuating between several points. what should I change to make the position of the robot stable? Many thanks and regards.

Running robot_pose_ekf and amcl

February 27, 2015, 4:28 pm

≫ Next: amcl warning....No laser scan received

≪ Previous: turtlebot position fluctuates

$

0

0

We are trying to use the [Leg Detector](http://wiki.ros.org/leg_detector) package on the Turtlebot to detect people and then send goals to the navigation stack to follow a person. Separately, running leg detection and amcl work fine. However, when you run them together there are problems. robot_pose_ekf, which the leg_detector needs, seems to interfere with amcl somehow. Can anyone explain why this happens and how I can get around it? I think it's problem some duplication in pose estimation, so maybe hopefully it's something easy that I'm missing. Thanks, Sarah