Foundations of Artificial Intelligence (FAI) Group
divider line


   Home    People    Research    Projects    Software    Teaching    Publications

About this page. Much of the research in FAI involves the development of software as part of the work. Our philosophy is to share the source code as much as possible. This page gives links to most of our developments, so far as ready for release. All downloadable material is free for research and educational purposes.

Categories of software. In planning, a quintessential part of the software development are planning tools, or planners, which solve planning problems. We furthermore construct benchmarks, example planning tasks/domains used for evaluating the performance of our tools. There finally is a growing "Other" category of tools developed in more recent branches of FAI research, pertaining to language generation, action policy learning with neural networks, analysis of such policies through model checking and visualization methods.

For planning tools/planners, most of our implementations are based on the Fast Downward (FD) platform, which has become the canonical implementation basis for classical AI Planning research. Our implementations on other bases are much smaller in number, and are mostly older developments; we therefore group these together in a single section below.


Neural Action Policy Verification and Testing.


Explainable Planning.


Minecraft instruction generation.

We have established a software infrastructure for running online user experiments with instruction giving in Minecraft.


Planning Benchmarks.


FD Partial Delete Relaxation.


FD Probabilistic Planning.


FD Star-Topology Decoupling.


Symbolic Search Library.


FD Abstractions & Symmetries. To get the following code please contact Michael Katz.


Planning Tools not based on Fast Downward.

Imprint
Data protection