Using RTAB Map just for obstacle detection

[jorgemia]

Hi,

I'm trying to get a robot to drive in an orchard outdoors. I'm currently using reactive navigation with a stereo camera (processing sensor data in real-time to navigate down the orchard) not SLAM, to navigate down a row. I was wondering if I could use the /rtabmap/cloud_obstacles from RTABMAP to just detect obstacles in the environment, not for navigation? I would like to use it to tell the robot, if there is an obstacle in front of you, at less than 2m, then stop.

Could this be done using RTABMAP? Not sure how I would then use the pointcloud to actually tell the robot to stop.

Is there a better approach?

[matlabbe]

Hi,

You can use obstacles_detection nodelet alone. What you are trying to do is very similar to what is done in this tutorial to give the obstacles to local costmap, which is reproduced in this test_obstacles_detection.launch file. Here this disparity image is taken from the output of stereo_image_proc node. If your camera gives already a depth image, you can use also the depth image with the point_cloud_xyz nodelet by remapping depth/image input topic instead.

cheers,
Mathieu