Use KnowRob from your program

The interactive Prolog shell that rosprolog provides is good for exploring KnowRob, visualizing knowledge, developing new functions and debugging Prolog code. However, if you would like to use KnowRob in your robot's control program, you need a way to send queries from your program. This functionality is provided by the json_prolog package. It provides a service that exposes a Prolog shell via ROS. You can run the json_prolog service using a launch file such as the following (which can be found in knowrob_map_data/launch/ccrl2_semantic_map.launch).

<launch>
  <param name="initial_package" type="string" value="knowrob_map_data" />
  <param name="initial_goal" type="string" value="owl_parse('package://knowrob_map_data/owl/ccrl2_semantic_map.owl')" />
 
  <node name="json_prolog" pkg="json_prolog" type="json_prolog_node" cwd="node" output="screen" />
</launch>

The json_prolog_node reads two optional ROS parameters for the initial package to be loaded (that you also give as argument when starting rosprolog) and a command to be executed at startup, for example for parsing an OWL file.

The communication with the json_prolog service uses a JSON-encoded format for representing the Prolog terms that form the query. While you can construct these terms manually and call the service directly, it is usally much easier to use one of the provided client libraries. There are libraries for Python, C++, Java and Lisp. Example clients are available in the json_prolog/examples folder; below is the code for sending KnowRob queries from Python via json_prolog.

#!/usr/bin/env python
 
import roslib; roslib.load_manifest('json_prolog')
 
import rospy
import json_prolog
 
if __name__ == '__main__':
    rospy.init_node('test_json_prolog')
    prolog = json_prolog.Prolog()
    query = prolog.query("member(A, [1, 2, 3, 4]), B = ['x', A]")
    for solution in query.solutions():
        print 'Found solution. A = %s, B = %s' % (solution['A'], solution['B'])
    query.finish()

#include <string>
#include <iostream>
 
#include <ros/ros.h>
#include <json_prolog/prolog.h>
 
using namespace std;
using namespace json_prolog;
 
int main(int argc, char *argv[])
{
  ros::init(argc, argv, "test_json_prolog");
 
  Prolog pl;
 
  PrologQueryProxy bdgs = pl.query("member(A, [1, 2, 3, 4]), B = ['x', A], C = foo(bar, A, B)");
 
  for(PrologQueryProxy::iterator it=bdgs.begin();
      it != bdgs.end(); it++)
  {
    PrologBindings bdg = *it;
    cout << "Found solution: " << (bool)(it == bdgs.end()) << endl;
    cout << "A = "<< bdg["A"] << endl;
    cout << "B = " << bdg["B"] << endl;
    cout << "C = " << bdg["C"] << endl;
  }
  return 0;
}

package org.knowrob.json_prolog;
 
import org.knowrob.json_prolog.Prolog;
import org.knowrob.json_prolog.PrologBindings;
import org.knowrob.json_prolog.PrologQueryProxy;
import org.knowrob.utils.ros.RosUtilities;
 
 
public class JSONPrologTestClient {
 
    public static void main(String args[]) {
 
        Prolog pl = new Prolog();
        RosUtilities.runRosjavaNode(pl, new String[]{"org.knowrob.json_prolog.Prolog"});
 
        PrologQueryProxy bdgs = pl.query("member(A, [1, 2, 3, 4]), B = ['x', A], C = foo(bar, A, B)");
 
        for(PrologBindings bdg : bdgs) {
            System.out.println("Found solution: ");
            System.out.println("A = " + bdg.getBdgs_().get("A") );
            System.out.println("B = " + bdg.getBdgs_().get("B") );
            System.out.println("C = " + bdg.getBdgs_().get("C") );
        }
  }
}