Efficient Search with an Ensemble of Heuristics
Authors: Mike Phillips, Venkatraman Narayanan, Sandip Aine, Maxim Likhachev
IJCAI 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We provide a theoretical analysis and compare our new strategies with the round-robin method on a 12-DOF full-body motion planning problem and on sliding tile puzzle problems. In these experiments, we used up to 20 heuristics and observed a several times speedup without loss in solution quality. |
| Researcher Affiliation | Academia | Mike Phillips Carnegie Mellon University Venkatraman Narayanan Carnegie Mellon University Sandip Aine Indraprastha Institute of Information Technology Delhi Maxim Likhachev Carnegie Mellon University |
| Pseudocode | Yes | Algorithm 1 MHA* Algorithm 2 Round-Robin Algorithm 3 Meta-A* Algorithm 4 DTS |
| Open Source Code | No | The paper does not provide any concrete access to source code, such as a specific repository link, an explicit code release statement, or a mention of code in supplementary materials for the methodology described. |
| Open Datasets | No | We used 100 randomly generated instances of puzzles for each size as our benchmark suite. For a given puzzle size, we generate a database of 1000 different solved configurations by performing a random walk of k (a random number between 2 and 10 times the puzzles size) steps from sgoal. |
| Dataset Splits | No | The paper does not provide specific dataset split information (exact percentages, sample counts, citations to predefined splits, or detailed splitting methodology) needed to reproduce the data partitioning. |
| Hardware Specification | No | The paper does not provide specific hardware details (exact GPU/CPU models, processor types, memory amounts, or detailed computer specifications) used for running its experiments. It only mentions the PR2 robot as the domain for motion planning, not as the computational hardware. |
| Software Dependencies | No | The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiment. |
| Experiment Setup | Yes | We ran SMHA* with DTS (C = 10), Meta-A* (wm = 10), and round-robin on all 100 trials. All three used wh = 25 and wa = 4. For this domain, we ran DTS with C = 1000 and Meta-A* with wm = 100. We also ran WA* (without re-expansions) on the same set of problems with h0 as the heuristic and w = 10. In contrast, MHA* (with wa = 2, wh = 5, bound = wa wh = 10) performs better, showing the benefit of multiple heuristics. |