| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
I need information on using RGB D-SLAM for obstacle avoidance in turtlebot. If I launch kinect+rgbdslam.launch in one terminal after launching openni.launch in different terminal, if I use rostopic pub cmd_vel for moving turtlebot, will it traverse detecting obstacle and avoiding it ? As obstacle avoidance algorithm is already implemented in RGBD-SLAM ?
No, RGBD SLAM only provides localization and mapping. That is, it tries to build a map and figure out where the robot is within that map. What you're talking about is path planning. If you look at the diagram on this page, it's the move_base package that does that planning.
I think this would be a good question to read: http://answers.ros.org/question/9712/how-to-use-rgbd-6d-slam-for-path-planning-and-navigation-with-kinect/.
-Jon
Hi Jon, Thanks a lot for the information. I have gone through the tutorials related to setting up and configuring navigation stack for robot. I just want to put it down here, please let me know if I am missing something here: I will be creating a separate package specific to the robot I am working
on , then create a robot configuration file and make changes to the launch file based on sensor nodes my robot makes use of and sensor_msgs type (point cloud for kinect etc.) , build it and then try running the launch file for my robot navigation stack. This is how I should proceed ?
I think so, pretty much. I don't know what you've done already, but it makes sense to first make sure you can drive the robot around. Then make sure you get the robot working with RVIZ. Then make sure you can get the laser scan working in RVIZ. Then gmapping. Then the autonomous navigation.
If you jump right into the navigation stack, it will be difficult to debug any problems.
Hi Jon, I have tried all the steps (laser scanning (since I am not using planar laser instead using Kinect, hence I have done the point_cloud to laser package for laser scanning, TF and publishing odometry ) and got all of them working on RVIZ. Hence, I think I can now get SLAM working now through
Hi Jon, For SLAM to work with kinect, I have created three launch files as follows --> one for kinect_laser to transform point cloud data to laser scan data, one for link between kinect and base_link (static_transform_publisher) and one for slam gmapping with all the required parameters.
And everything worked fine and I could save the map using map_saver. I think I can now do the autonomous navigation of this known map. Thanks a lot for the help and links for 2D SLAM with kinect.
Hi, I started working on rgbd slam with TurtleBot, ros hydro Ubunto 12.04 I followed thus link:http://felixendres.github.io/rgbdslam_v2/ to install rgbd slam
and after I had a GUI window as the window exactly in this link, with the two posts below: waiting for visual picture on topic and waiting for depth image is topic there is no picture, i think it was a problem with topics in launch files can some one help me and send to me the launch of openni+rgbdslam.launch openni.launch rgbdslam.launch
Hi I'm using Occipital (RGB-D sensor like Kinect) on Parrot AR. DRONE for navigation using SLAM algorithm.... What do I do for obstacle avoidance, is it done by the SLAM or should I use additional sensors If I use additional sensors, how do I make it a part of the SLAM algorithm.
This is not an answer to the original question. Please create a new question specific to your use-case and do not post questions as answers, as this is confusing to other readers.
Asked: 2013-03-17 02:05:20 -0500
Seen: 6,409 times
Last updated: Mar 26 '15
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
