Red-Black Heuristics for Planning Tasks with Conditional Effects
Authors: Michael Katz7619-7626
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We show empirically that red-black planning heuristics that handle conditional effects natively outperform the variants that compile this feature away, improving coverage on tasks where black variables exist by 19%. ... Experimental Evaluation In order to evaluate the benefit of natively supporting conditional effects in red-black planning heuristics, we adapted the existing implementation of red-black planning heuristics on top of the current Fast Downward framework (Helmert 2006). |
| Researcher Affiliation | Industry | Michael Katz IBM Research Yorktown Heights, NY, USA michael.katz1@ibm.com |
| Pseudocode | Yes | Figure 2: Red-black planning algorithm. |
| Open Source Code | No | The paper states 'we adapted the existing implementation of red-black planning heuristics on top of the current Fast Downward framework (Helmert 2006)', but does not provide concrete access to their specific implementation/code for the described methodology. |
| Open Datasets | Yes | We performed our evaluation on the existing set of benchmark domains with conditional effects. ... following Haslum (2013), we also use problems generated by the conformant-to-classical planning compilation (T0) (Palacios and Geffner 2009) and the finite-state controller synthesis compilation (FSC) (Bonet, Palacios, and Geffner 2009). In addition, following R oger, Pommerening, and Helmert (2014), we use the briefcase world domain from the IPP benchmark collection (K ohler 1999) and the Miconic simpleadl version from the benchmark set of the International Planning Competition (IPC2000), as it has conditional effects but no derived predicates after grounding with Fast Downward. Finally, we use the domains from the most recent IPC 2018. |
| Dataset Splits | No | The paper uses existing benchmark domains for evaluation. It does not explicitly state specific training, validation, or test dataset splits (e.g., percentages, sample counts, or a detailed splitting methodology). |
| Hardware Specification | Yes | The experiments were performed on Intel(R) Xeon(R) CPU E7-8837 @2.67GHz machines, with the time and memory limit of 30min and 2GB, respectively. |
| Software Dependencies | No | The paper mentions 'Fast Downward framework (Helmert 2006)' but does not provide specific version numbers for this or any other software dependencies. |
| Experiment Setup | Yes | We perform a greedy best first search with a single queue, ordered by the red-black heuristic. ... The experiments were performed on Intel(R) Xeon(R) CPU E7-8837 @2.67GHz machines, with the time and memory limit of 30min and 2GB, respectively. |