Optimising Partial-Order Plans Via Action Reinstantiation

Authors: Max Waters, Lin Padgham, Sebastian Sardina

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We also propose a MAXSAT-based technique for increasing plan flexibility and provide a thorough experimental evaluation that suggests that there are benefits in action reinstantiation.
Researcher Affiliation Academia Max Waters , Lin Padgham and Sebastian Sardina RMIT University, Melbourne, Victoria 3000, Australia {max.waters, lin.padgham, sebastian.sardina}@rmit.edu.au
Pseudocode No The paper describes encoding methods using formulas (e.g., Formulae 1-17) but does not present them in a structured pseudocode or algorithm block.
Open Source Code Yes Implementation is available at bitbucket.org/max waters/mrr
Open Datasets Yes Test cases (i.e., input plans) were generated by giving all first-order IPC STRIPS planning instances to three planners.
Dataset Splits No The paper generates 'test cases' (input plans) from IPC STRIPS planning instances and evaluates optimization techniques, but it does not describe specific training, validation, or test splits for a dataset in the context of model training.
Hardware Specification Yes Plan generation and encoding/optimisation were both limited to 8GB and 30m at 2.60GHz.
Software Dependencies No The paper mentions 'Loandra MAXSAT solver' and 'Max Pre [Korhonen et al., 2017]' but does not provide specific version numbers for these software components.
Experiment Setup Yes Test cases (i.e., input plans) were generated by giving all first-order IPC STRIPS planning instances to three planners. To ensure a variety of plans, three planners of distinct types were used: the novelty-driven best-first search planner Dual BFWS [Lipovetzky and Geffner, 2017], the heuristic forward-search planner LAMA [Richter and Westphal, 2010] and the SAT planner Madagascar [Rintanen, 2010]. Each (unique) plan was encoded with each encoding2, and the resulting MAXSAT instances were preprocessed with Max Pre [Korhonen et al., 2017] and given to the Loandra MAXSAT solver. Plan generation and encoding/optimisation were both limited to 8GB and 30m at 2.60GHz.