Home

ROSPlan_interface_Turtlebot2

This package makes the robot to rotate around quickly so that the laser scans (white lines created in Rviz when robot is facing an obstacle) can be matched to the map in order to localise the vehicle. Instead of manually giving a pose estimate of the robot in RViz, this component will be able to "localise" the robot in current map; so that pose of robot in /base_link will match actual location in /map.

RPLocaliser.cpp file retrieves waypoint parameters from "waypoint.txt file in rosplan_demos/scripts directory. Make sure you create this .txt file including two random waypoints, such as:

wp_name [0, 0, 0.234] 
wp_name2 [5, 4, 0.234] 

This parameter file needs to be included in interfaced_planning_system.launch file by adding the following command:

<param name="waypoint_file" value = "$(find rosplan_demos)/scripts/waypoints.txt" />

Make sure that this package node is included in the same launch file:

<node name="rosplan_interface_localise" pkg="rosplan_interface_turtlebot" type="rplocaliser" respawn="false" output="screen"> 
</node> 

It's desired to perform this action before visiting any waypoints. Therefore, the PDDL action below, called "localise", is added to the domain, and a propositional condition of (localised ?v) is added to existing goto_waypoint action.

 (:durative-action localise 
  :parameters (?v - robot) 
  :duration ( = ?duration 10) 
  :effect (and (at end (localised ?v))))) 

When turtlebot_explore.bash file is launched, planning system will generate a valid plan (using POPF), like:

0.000: (localise kenny)  [10.000] 
10.001: (goto_waypoint kenny wp0 wp0)  [10.000] 
20.002: (goto_waypoint kenny wp0 wp1)  [10.000] 
30.003: (goto_waypoint kenny wp1 wp2)  [10.000] 

And, it can be observed that Planning System (PS) dispatches "localise

---

" action, and is received by rosplan_interface_localise, while robot is turning around to localise itself in the simulation:

KCL: (PS) Dispatching action [0, localise, 10.037968, 10.000000] KCL: (Localiser) action received

You can also observe that AMCL particle cloud gets tightened after localisation, because robot gets a better pose estimation after this action.

---