[Chair]  [Computer Science Department (FBI)]  [University Dortmund] 

Motivation

Introduction

Architecture

Grounding

SAS+ Encoding

Goal Ordering

Operator Dependency

Symmetry

Domain Inference

Planning

Validation and Visualization

ModPlan - Modern Action Planning


The MopPlan workbench is an interacive knowledge acquisition and engineering tool for AI planning. It provides automated domain analysis tools together with PDDL learning capabilities. Integrated optimal and suboptimal planning technology extends state-of-the-art technology.

With the tool, domain experts assist solving hard combinatorial problems. Approximate or incremental solutions provided by the system are supervised. Intermediate results are accessible to improve domain modeling and to tune exploration in generating enhanced plans, which, in turn, can be bootstrapped for domain description inference.


Download (Windows Execs):

Tar-File

Directory

ModPlan

ModPlan is being developed at the University of Dortmund within a student project group supervised by Stefan Edelkamp, and Tilman Mehler

Thanks to Edina Kurtic, Miriam Bützken, Andrea Matuszewski, Rachid Karmouni, Michael Nelskamp, Arne Wiggers, Abdelaziz Elalaoui, Khalid Lahiane, Mohammed Nazih, Kenneth Kahl, Roman Klinger, and Abubakr Mkhdramine for all their implementation efforts.

Moreover, we are grateful to Björn Scholz and Shahid Jabbar for improving the vizualization of plans and to Malte Helmert for providing access to translate.

The work is supported by DFG in the projects Ed 74/2 and 74/3.