detect dynamic obstacles with Kinect and RTAB-MAP

[VEGETA]

Hi there friends,

I want to make a robot (mobile robot) that does SLAM in 3D, I've seen that your package is the best. what I aim for is:

1- use kinect to get map data.
2- make a 3d map with all necessary features for navigation.
3- use multi agents having kinect to make the map... like putting kinect devices in the room to cover all the area, and generate one map of all that.
4- detect dynamic obstacles... this is the most important one.
5- use ros in all above, PLUS make a planner to adjust robot's plan according to the dynamic obstacle's movement trajectory.

I don't want to use other types of sensors especially laser ones. it will all be visual via kinect.

please guide me in full details... is this a good and achievable idea? can it be done without years of coding (I am a bad programmer)? should I do anything else?

thanks!

[matlabbe]

Hi,

I would recommend to start by doing navigation in 2D with a single robot. rtabmap can provide a global map (occupancy grid map) for the navigation stack of ROS: http://wiki.ros.org/navigation/Tutorials/RobotSetup

For 4 and 5, the navigation stack does that.

A full integrated example rtabmap + navigation stack can be found here with TurtleBot: http://wiki.ros.org/rtabmap_ros/Tutorials/MappingAndNavigationOnTurtlebot

If your robot is differential, it should be similar. Browse the referred launch files and linked tutorials.

cheers

hi vegeta,

i am doing the same project and i am unsure what to do.have u been able to come up with something?