Visualizing costmaps in the opengl window

[gaya-]

Right now we have global variables for the Z and the orientation. See b8ae8b4

Would be nice to be able to specify those during costmap generation not as a global variable for all the costmaps.

[fairlight1337]

Was this change done out of practicability reasons? Looks like the height was there already (z) :). Maybe we can extend that?

[gaya-]

It is there in the low-level API but gets lost in the high-level somewhere. I didn't look much into this, just opened this issue not to forget about it.

[fairlight1337]

Hm, thinking about it again, doesn't the OpenGL visualization already show the costmaps on the floor or a table, depending on the z-value generator within the location generators? Such as make-semantic-map-height-function in semantic_map_costmap (https://github.com/cram-code/cram_highlevel/blob/master/semantic_map_costmap/src/cost-functions.lisp#L338)?

My OpenGL visualization draws locations for robot standing poses on the floor, and for locations on table, it draws them on the table. The height is decided based on the generators registered for those location designators through the respective (<- (desig-costmap ...)) predicates.

What exactly seems to be the problem then?

[gaya-]

No, unfortunately, it is always generated on the floor. The reason why it seems to be on the table when we request locations for objects is that the values of the costmap are relatively high (if the distribution is uniform it's 1.0, otherwise it's usually still close to 1.0), and the table height is about 0.8, so it seems like it's on the table. But if the distribution would not have such high values you wouldn't be able to see some parts of it because they would be inside the table. So, basically, the reds of the robot pose distributions are on the same height as the reds of the object distributions. So it isn't connected to the height generators. Which is exactly what I would prefer to have.

The point where the visualization is created is here: https://github.com/cram-code/cram_physics/blob/master/bullet_reasoning_designators/src/location-designator-integration.lisp#L61 and there the Z value is left on the default of 0.0.

[fairlight1337]

Oh, I see. Okay then, haven't had any issues with that.