2. Run the Human Aware Navigation

If you want to test the social navigation behavior in simulation, first start the gazebo simulator with turtlebot(3) robot:

$roslaunch human_aware_nav_launch sim_world.launch

Otherwise, start the robot and the robot driver. To start the human aware navigation, launch:

$roslaunch human_aware_nav_launch navigation.launch

This command starts move_base with the lattice_planner as the global planner for time dependent planning and the timed_path_follower as the local planner for path following. Please refer to their GitHub pages for more information of how to configure them. The layered dynamic cost map plugin for the lattice_planner is specified as the social layers plugin from the dynamic_social_costmap package (from the human_aware_navigation repository). Furthermore, the launch file starts a constant velocity prediction node and rviz.

You should already be able to set goals in rviz and watch the robot navigate autonomously. To test the behavior around humans you can run a fake prediction node that simulates a person walking in a straight line with constant velocity from a starting position:

$rosrun people_detection fake_detection

To change the starting position, run the command with parameters and change them to your desired velocity and starting position (in map coordinates):

$rosrun people_detection fake_detection _x:=-5.74 _y:=-1.1 _phi:=0.0 _vel:=0.4

Of course you can also fire up your favourite people detector to detect people in the environment. Make sure it publishes people_msgs::People messages on the "people" topic with velocity information.