Write a GPS “problem” for the following scenario. Make sure GPS can correctly find a plan. See below for how to code and execute GPS.
The scenario: Your sister wants to go to a concert. Your brother wants to go to the park. You are at home, your sister is at the park, and your brother is at home. You have a car, and can drive to the various places. You can carry one passenger who has to be “in” the car. The pickup action should put the person in the car, and the dropoff action should take them out of the car. At the end, you want to be back home. What is a plan that meets the goal?
Download paip.zip, which contains Python code for GPS. This code comes from Daniel Connelly, who adapted the code from Peter Norvig’s book Paradigms of Artificial Intelligence Programming: Case Studies in Common Lisp.
Edit the file ~paip/examples/gps/hw7.py~ It looks like this:
from paip.gps import gps
problem = {
"start": [],
"finish": [],
"ops": []
}
def main():
start = problem['start']
finish = problem['finish']
ops = problem['ops']
for action in gps(start, finish, ops):
print action
if __name__ == '__main__':
main()Don’t change def main() and anything below that. Change the
definition of the problem variable. Look at the other examples in
that folder for guidance, as well as our GPS notes.
Run GPS like this: Open a command prompt or terminal. cd into the
directory with the run_examples.py file. Type python
run_examples.py, and you should see this list. (Note, this requires
Python 2.7).
Please choose an example to run: 0 paip.examples.eliza.eliza 1 paip.examples.eliza.support 2 paip.examples.emycin.mycin 3 paip.examples.gps.blocks 4 paip.examples.gps.door 5 paip.examples.gps.eat 6 paip.examples.gps.hw7 7 paip.examples.gps.monkeys 8 paip.examples.gps.school 9 paip.examples.logic.find_elements 10 paip.examples.logic.find_length 11 paip.examples.logic.find_list 12 paip.examples.logic.find_lists_lengths 13 paip.examples.logic.find_list_length_4 14 paip.examples.logic.likes 15 paip.examples.logic.transitive 16 paip.examples.othello.othello 17 paip.examples.search.gps 18 paip.examples.search.pathfinding >> 6
Type 6 to run your HW7 code. It should find a correct plan. Try the
other GPS examples to see what you should expect. (Ignore the “None”
at the start of each plan.)
Turn in ~hw7.py~.
Write a new solution to the same scenario in PDDL format. This
format requires two files: a “domain” file and a “problem” file. Name
your domain file hw7-domain.pddl and your problem file
hw7-prob.pddl. They should start like this (replace the “…”):
;; hw7-domain.pddl
(define (domain hw7)
(:requirements :strips)
(:constants car self)
(:predicates (at ?x ?y)
...)
(:action goto
:parameters (?from ?to)
:precondition (and (at car ?from)
(in self car))
:effect ...)
...)The problem file starts like this:
;; hw7-prob.pddl
(define (problem hw7-prob)
(:domain hw7)
(:objects car self sister brother home concerthall park)
(:init (at car home) ...)
(:goal (and (at sister concerthall)
...)))Test your domain/problem with JavaFF.zip (from: Coles et al.). Download
the ZIP and put your hw7-domain.pddl and hw7-prob.pddl files in
the extracted folder. Run like this:
java -cp . javaff.JavaFF hw7-domain.pddl hw7-prob.pddl
You should also be able to run the Cargo example, just to be sure you downloaded everything correctly:
java -cp . javaff.JavaFF cargo-domain.pddl cargo-prob.pddl C:\Users\Joshua\Documents\teaching\cse3521\javaff>java -cp . javaff.JavaFF cargo-domain.pddl cargo-prob.pddl Parsed Domain file cargo-domain.pddl successfully Parsed Problem file cargo-prob.pddl successfully Performing search as in FF - first considering EHC with only helpful actions 6 4 3 1 (load c2 p1 jfk) (fly p1 jfk sfo) (unload c2 p1 sfo) (load c1 p2 sfo) (fly p2 sfo jfk) (unload c1 p2 jfk) Instantiation Time = 0.073sec Planning Time = 0.071sec
Turn in ~hw7-domain.pddl~ and ~hw7-prob.pddl~.