SoCS 2022: AI Planning Benchmarks and Experiment Data for Decoupled Search

Silvan Sievers from the University of Basel provides code, scripts, and data for reproducing experiments from the SoCS 2022 paper on additive pattern databases. The bundle includes implementations based on Fast Downward, experiment scripts for Lab 7.0, and benchmark suites from IPC and Autoscale. Raw and processed experimental data from planner runs are also included.

Use Cases

• Reproducing planning algorithm experiments based on the provided Lab 7.0 scripts and configurations.
• Benchmarking new planning methods against established IPC and Autoscale benchmark suites included in the data.
• Analyzing planner performance based on the raw log files and parsed properties files from individual runs.

Strengths

• Includes raw and processed experimental data from planner runs, with structured properties files.
• Bundles multiple established benchmark suites, specifically the STRIPS IPC benchmarks up to 2018 and Autoscale 21.11 benchmarks.
• Provides a complete reproduction package with code, scripts, and a copy of the Lab 7.0 experiment framework.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

University of Basel, associated with the SoCS 2022 paper.

Collection Method

Data generated from experiments reported in the academic paper, using planner runs on benchmark problems.