Revision history [back]

I saw also this question from you; looks good but you have the infrared scan in the same z_voxel as the kinect (min/max_obstacle_height are the same for both). You should have z_voxels = 3, and use one for bumpers, one for infrared and one for the kinect, ensuring tat min/max_obstacle_height values lay within the z-voxel limits.

In my case I use a ring of sonars instead of infrared, but the integration with navi stack is the same. However, I publish the sonar readings as sensor/msgs/Range and use range_sensor_layer to add them to the costmap. You don't need to do this, really; if you are using typical Sharp IR sensors, that have very narrow fov and are not very noisy, merge them in a laser scan is fine.