ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question

Mapping with Kinect sensor

asked 2014-08-11 04:24:34 -0500

sumanth gravatar image

I am working on a robot, which finally should move to the specified location in the map autonomously. (For which I am planning to use Navigation stack).

For which I want to map the environment. I have a kinect sensor available, so which mapping I should use so that I can generate a map (2D or 3D), which can be later fed to the navigation stack....??

I can fetch the odo ticks from the micro controller and give for the mapping stack. As the kinect is a fixed connection to the robot, I can publish a static transform from kinect to the base link.

Are these sufficient to use the mapping stack, but the question is which mapping stack to go with..? and any examples suggesting the use of the mapping stack will be great full to start with.

Many thanks in advance.

edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted

answered 2014-08-11 05:17:54 -0500

bvbdort gravatar image

For the task of reaching goal 2d map is sufficient.

First generate the map by moving robot manually . Use gmapping with 2d laser and odometry. (use depthimage_to_laserscan to generate laser scan from kinect data)

When you have map use amcl for localisation. This tutorial of Autonomous Navigation of a Known Map with TurtleBot is helpful.

edit flag offensive delete link more


Thanks, but then where the 3D map is used..?

sumanth gravatar image sumanth  ( 2014-08-11 06:54:24 -0500 )edit

check the videos here . Of course you can use 3d maps for navigation. try rgdb slam but i never get chance to use 3d navigation.

bvbdort gravatar image bvbdort  ( 2014-08-11 08:03:19 -0500 )edit

but does navigation stack takes the map from the gmapping.?

sumanth gravatar image sumanth  ( 2014-08-12 01:59:25 -0500 )edit

Are there any exampels/tutorial to generate the map by moving robot manually . Use gmapping with 2d laser and odometry..?

sumanth gravatar image sumanth  ( 2014-08-12 02:14:11 -0500 )edit

first you need to build the map using gmapping, then use map for navigation. Which robot are you using ? and how you are driving it ß

bvbdort gravatar image bvbdort  ( 2014-08-12 02:15:13 -0500 )edit

I am using my custom built robot, its a differential drive mobile robot, plan to drive the robot for mapping is using commands from laptop (similar to teleop on turtlebot).

sumanth gravatar image sumanth  ( 2014-08-12 02:27:47 -0500 )edit

start with building map by driving robot

bvbdort gravatar image bvbdort  ( 2014-08-12 03:21:06 -0500 )edit

@bvbdort, I tried mappin with the gmapping but was not sucessful. Steps followed:

  1. open the kinect with openini using "roslaunch openni_launch openni.launch".
  2. pointcloud to the laser scan using "rosrun depthimage_to_laserscan depthimage_to_laserscan image:=/camera/depth/image_raw"
sumanth gravatar image sumanth  ( 2014-08-13 02:43:56 -0500 )edit

Question Tools

1 follower


Asked: 2014-08-11 04:24:34 -0500

Seen: 5,644 times

Last updated: Aug 11 '14