This repository contains the code for the paper:

Asking the Right Questions: Learning Interpretable Action Models Through Query Answering.
Pulkit Verma, Shashank Rao Marpally, and Siddharth Srivastava.
35th AAAI Conference on Artificial Intelligence, 2021.

Paper | Extended Version | Talk | Slides | Poster

|-- dependencies/
|   |-- FD/
|   |-- FF/
|   |-- pddlgym/
|   |-- VAL/
|-- domains/
|-- random_states/
|-- results/
|-- src/
|   |-- agent.py
|   |-- config.py
|   |-- generate_random_states.py
|   |-- main.py
|   |-- interrogation/
|   |-- lattice/
|   |-- query/
|   |-- sim/
|   |-- utils/
|-- README.md
|-- LICENSE

• dependencies: This directory includes the external software used to run the code. This includes FF, FD, VAL, and PDDLGym.

  • FF: https://fai.cs.uni-saarland.de/hoffmann/ff/FF-v2.3.tgz
  • FD: https://github.com/aibasel/downward
  • PDDLGym: https://github.com/tomsilver/pddlgym
  • VAL: https://github.com/KCL-Planning/VAL

• dependencies: Place all the domains in this directory. There must be a directory for each domain containing:

  • domain.pddl (domain file for that domain), and
  • instances directory containing the problem files for that domain.

• random_states: This directory stores the set of states in serialized form. For each domain, there is a .pkl file containing 60 states approximately. These are generated using src/generate_random_states.py.

• src: This directory stores the source code for AIA. It contains 4 files:

  • agent.py: Contains the agent code.
  • config.py: Declares the configuration parameters.
  • generate_random_states.py: Generates the random states for each domain.
  • main.py : Contains the main driver code which runs the code end-to-end.

  src also contains code structured into following sub-directories:

  • interrogation: Contains the AIA code.
  • lattice: Contains the model and lattice classes.
  • query: Contains the plan outcome query code.
  • sim: Simulator specific code. Contains a separate agent file for each simulator domain.
  • utils: Contains general utilities.
    • utils/parser: Code based on PDDLGym.
    • utils/translate: Code based on FD.

Configuration parameters are set in src/config.py

• FF_PATH, FD_PATH, and VAL_PATH stores the relative path of FF, FD, and VAL respectively.
• NUM_PER_DOMAIN denotes how many instances per domain must be run. Keep minimum 2 for symbolic agent.
• PLANNER specifies which planner to use. Set it to either FF or FD.
• Comment out either Symbolic Agent Settings or Simulator Agent Settings.

1. Install the required software

pip3 install -r requirements.txt 

2. Adjust variables/paramters as needed in src/config.py.

3. Run main.py

cd src
python3 main.py

1. OpenCV (Tested on Ubuntu 18.04)

   Refer to https://linuxize.com/post/how-to-install-opencv-on-ubuntu-18-04/

2. FF:

   • Please install flex and bison for FF to compile.

   • On newer versions of gcc (tested on gcc 10.2.0) please make the following changes:

     • main.c:150 : Comment out the gbracket_count definition

       int gbracket_count; --> /* int gbracket_count; */
       

     • relax.c:111 : Define lcurrent_goals as static

       State lcurrent_goals; --> static State lcurrent_goals;
       

     • search.c:110 : Define lcurrent_goals as static

       State lcurrent_goals; --> static State lcurrent_goals;
       

Please note that this is research code and not yet ready for public delivery, hence most parts are not documented.

In case of any queries, please contact verma.pulkit@asu.edu, or smarpall@asu.edu.

Pulkit Verma
Shashank Rao Marpally
Abhyudaya Srinet
Siddharth Srivastava

@inproceedings{verma_2021_asking,
    author = {Verma, Pulkit and Marpally, Shashank Rao and Srivastava, Siddharth},
    title = {{Asking the Right Questions: Learning Interpretable Action Models Through Query Answering}},
    booktitle = {Thirty-Fifth AAAI Conference on Artificial Intelligence (AAAI-21)},
    year={2021}
}