Gmapping from bagfile- KinectV1 to scan- bad results

asked 2021-02-15 16:34:33 -0500

AGV5 gravatar image

Hi ,

I am currently working on a school project where a mobile robot (Festo Robotino 3 basic) need to navigate through the lab. We decided to go with ROS and have the following setup:

Robot Robotino: (Client)

intel atom


  • Ubuntu 18.04
  • ROS Melodic


  • freenect_launch
  • pointcloud_to_laserscan

Computer: (Master)

intel i3


  • Ubuntu 18.04
  • ROS Melodic


  • Turtlebot3_SLAM
  • robotino_node (uses the API of the robot to publish /odom and subscribe to /cmd_vel)

In order to work through suitable parameters to create a useful map. We did a couple round through the lab and created a bag file:

As mentioned the scan data comes from a kinect v1 (55degrees viewing angle and from each column the closest pixel is chosen). We have tried to create a 2DMap from the bagfile, the results where all not usful, mainly due to constant jumping of odom relative to the map frame (we tried to increase the minimumScore to 10000, it did not create a better result ) so the main question is: Would somebody be willing to check if they are able to create a somewhat useful 2D map from that bag file? And if not, is there something that we can address in order to obtain better data.

Thanks for your help.


map_update_interval: 3.0 
throttle_scans: 2   
maxUrange: 2.5
sigma: 0.05
kernelSize: 1
lstep: 0.05    
astep: 0.05
iterations: 5
lsigma: 0.075
ogain: 3.0
lskip: 1
minimumScore: 100
srr: 0.05
srt: 0.1
str: 0.05
stt: 0.1
linearUpdate: 0.2
angularUpdate: 0.1
temporalUpdate: 3
resampleThreshold: 3
particles: 300
xmin: -10.0
ymin: -10.0
xmax: 10.0
ymax: 10.0
llsamplerange: 0.01
llsamplestep: 0.01
lasamplerange: 0.005
lasamplestep: 0.005
maxRange: 2.49

strong text

edit retag flag offensive close merge delete