Top-Quality Planning: Finding Practically Useful Sets of Best Plans
Authors: Michael Katz, Shirin Sohrabi, Octavian Udrea9900-9907
AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We empirically demonstrate the superior performance of our approach compared to a top-k planner-based baseline, ranging from 41% increase in coverage for finding all optimal plans to 69% increase in coverage for finding all plans of quality up to 120% of optimal plan cost. 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. The benchmark set consists of all STRIPS benchmarks from optimal tracks of International Planning Competitions (IPC) 1998-2018, a total of 1797 tasks in 64 domains. Table 1 depicts per-domain coverage, comparing our technique, tq, to the baseline, K-tq, for four quality bound multipliers. |
| Researcher Affiliation | Industry | Michael Katz, Shirin Sohrabi, Octavian Udrea IBM T.J. Watson Research Center 1101 Kitchawan Rd, Yorktown Heights, NY 10598, USA |
| Pseudocode | Yes | Algorithm 1 Iterative unordered top-quality planning. Input: Planning task Π, quality bound q P , Π Π, π optimal plan to Π while cost(π) q do P P {π} {π |π is symmetric to π, π UΠ[π]} Π Π P , where Π P is the FMPMA reformulation π optimal plan to Π end while return P |
| Open Source Code | Yes | To evaluate the feasibility of our suggested approach, we have implemented an unordered top-quality planner as part of the Forbid Iterative planners collection (Katz, Sohrabi, and Udrea 2019), on top of the Fast Downward planning system (Helmert 2006). Katz, M.; Sohrabi, S.; and Udrea, O. 2019. Forbid Iterative planners for top-k, top-quality, and diverse planning problems. https://doi.org/10.5281/zenodo.3246773. |
| Open Datasets | Yes | The benchmark set consists of all STRIPS benchmarks from optimal tracks of International Planning Competitions (IPC) 1998-2018, a total of 1797 tasks in 64 domains. |
| Dataset Splits | No | The paper does not explicitly mention training/validation/test dataset splits. It describes evaluating on a benchmark set of tasks from International Planning Competitions. |
| 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 planning system (Helmert 2006)' and 'Forbid Iterative planners collection (Katz, Sohrabi, and Udrea 2019)' but does not provide specific version numbers for these or other software dependencies. |
| Experiment Setup | 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. We use Naive S, the best performing configuration of the iterative top-k planner (Katz et al. 2018), that exploits both symmetries and plan reorderings. We refer to our baseline as K-tq. The purpose of setting k to a large number is to allow the top-k planner to exploit the entire 30min time interval. However, once 10^6 plans were produced within the time bound, we stop. For each task, the quality bound is computed using the cost of the first found optimal plan, multiplied by a constant. We experiment with four different quality bound multipliers: qm = 1.0 (optimal plans only), 1.05, 1.1, and 1.2 of the optimal plan cost. |