Hello,
I've been trying to solve this problem since March with no luck. I posted [the issue](https://github.com/DLu/navigation_layers/issues/34) on the author's Github but it is still unresolved.
My environment is ROS Indigo (Debian install) on Ubuntu 14.04. I have installed the [social_navigation_layers](http://wiki.ros.org/social_navigation_layers?distro=indigo) package and the [people_msgs](http://wiki.ros.org/people_msgs) package and I have added the ProxemicLayer plugin to my costmap config files as follows:
plugins:
- {name: social_layer, type: "social_navigation_layers::ProxemicLayer"}
- {name: obstacle_layer, type: "costmap_2d::ObstacleLayer"}
- {name: inflation_layer, type: "costmap_2d::InflationLayer"}
I have verified that the plugin is loaded when move_base is launched and I see the layer in rqt_reconfigure where I can change its parameters or disable/enable it.
In the meantime, I have a person detector running on the camera stream on my robot. The person detector publishes detections on a topic (/people) using message type people_msgs/People. I have also used the static person publishers from the people_velocity_tracker package. In either case, the costmap in RViz in does not change whether the people are visible or not. For example, if I publish a static person using the [people_velocity_tracker](http://wiki.ros.org/people_velocity_tracker?distro=indigo) package, I would expect an obstacle to appear in the costmap located at the x-y coordinates specified in the /people topic message. But nothing appears in the costmap.
So I did some debugging by inserting some print statements in proxemic_layer.cpp. I have verified that the people_msgs/People topic is being subscribed to and the callback is getting the correct number of detected people and their x-y coordinates. However, the costmap does not appear to be updated.
Running rqt_console does not indicate any errors or exceptions.
Is anyone else using the social_navigation_layers package successfully?
Thanks!
patrick
patrick