Robotics StackExchange | Archived questions

Navigation not using Gmapping map, no obstacles appear

Hi my name is Cris and im a Chilean master student that is using ROS. Right now im stuck and can not continue with my thesis due to some problems with the gmapping and navigation stacks. I have been unable to advance for several week. Right now i can create correctly the map with gmapping but when i use the navigation stack the robot does not use this map and simply goes to the goal directly. Also when i look for obstacles non appear in rviz so i think thats the problem.

The first problem that appears is [ WARN] [1374781904.809778181]: The scan observation buffer has not been updated for 1.29 seconds, and it should be updated every 0.20 seconds.

And it continues forever

but when i change the transform_tolerance all this does is delay the warning, because when the time gets to the new parameter the warning appear again. When the warning dont appear i con give in rviz a location and a goal to the navigation stack and the robot ( a pioneer 3at) mover correctly to the goal but does not take in consideration the obstacles in his path. But when the warning appears again navigation doe snot work and even though i give a position and a goal the robot does not move.

Also when i do

rostopic echo /movebase/localcostmap/obstacles

I get header: seq: 1445 stamp: secs: 1374781469 nsecs: 703920572 frameid: /odom cellwidth: 0.0250000003725 cell_height: 0.0250000003725 cells: []

and cells is always [] , meaning it does not recognize a obstacle even though theres one in front of him.

Im using a pointcloudtolaserscan from the opennicamera, and works fine i can see the map and the /scan topics on rviz, but this might be the problem. Because men i rosrun tf viewframes

i gel map->/odom->/baselink->opennicamera and it does not end in /scan which is weird because slam is working fine

Im running this in ROS fuerte and any doubt ill give you the details, i would really appreciate some help

Asked by ctguell on 2013-07-25 10:00:58 UTC

Comments

What kind of sensor do you use to find obstacles ? Are you sure they're working well ?

Asked by Lucile on 2013-07-25 22:05:45 UTC

Im using a kinect and then I use a pointclout_to_laserscan to obtain a /scan . This is working good in gmapping because i can see the /scan in rviz but im not sure if it is correctly connected to the navigation stack, any help is welcome thanks

Asked by ctguell on 2013-07-26 08:24:11 UTC

Did you ever resolve this problem? I am about to try something similar.

Asked by gershon on 2013-08-29 09:08:40 UTC

Yes i had a really stupid problem with a variable

Asked by ctguell on 2013-08-29 09:36:51 UTC

Answers