PDSim Backend

Using Unified Planning Library to parse PDDL and generate plans for PDSim

Install

This project is now hosted on pypi

Use Conda (preferred)

Create a new venv

conda create -n pdsim python=3.11

Activate

conda activate pdsim

Install PDSim-Backend

pip install pdsim-backend

Use Venv

Create a new venv

python -m venv <directory>

Activate

• Windows(Powershell): <directory>\Scripts\activate.ps1
• Linux/macOS: source <directory>/bin/activate

Install PDSim-Backend

pip install pdsim-backend

Usage

Python Mode

from pdsim_backend import run_backend

run_backend(domain='./examples/pddl/blocks/domain.pddl', problem='./examples/pddl/blocks/problem.pddl')

CLI Mode

• Provide your domain and problem files.

python pdsim_unity.py --domain <domain_path> --problem <problem_path>

You can provide an optional --planner flag, by default it'll use fast-downward, but the user will be prompted which planner is available for a specific problem.

• Embed pdsim server in your up problem definition.

from pdsim_unity import pdsim_upf

< your  problem definition >

pdsim_upf(up_problem, planner_name)

This will create a server to communicate with unity and serve the protobuf representation of the problem and the generated plan.

PDSim will try to find the planner that suits best the planning problem provided and let you choose which one to run.

Some Availbale planners:

• FastDownward
• Tamer
• Pyperplan FUll list here