Avoiding Dead Ends in Real-Time Heuristic Search

Authors: Bence Cserna, William Doyle, Jordan Ramsdell, Wheeler Ruml

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We empirically test these new methods on two simulated domains, traffic crossing and controlling a vehicle with inertia, and find that the new methods dramatically outperform the conventional ones.
Researcher Affiliation Academia Bence Cserna, William J. Doyle, Jordan S. Ramsdell, Wheeler Ruml Department of Computer Science University of New Hampshire Durham, NH 03824 USA bence, doyle, ruml at cs.unh.edu
Pseudocode Yes Algorithm 1: LSS-LRTA*. and Algorithm 2: Safe RTS
Open Source Code No The paper implements algorithms in Kotlin and references the Kotlin programming language (Jet Brains 2017), but does not provide any explicit statement about releasing the source code for the methodology described in the paper, nor does it include a link to a code repository.
Open Datasets No The paper describes the creation of custom instances for its experiments, such as 'We created 100 random instances of 50 by 50 cells' for the traffic domain and 'we created 25 instances with starting positions chosen randomly' for the racetrack domain. However, it does not provide concrete access information (e.g., links, DOIs, repository names, or formal citations) for these datasets to be publicly available or open.
Dataset Splits No The information is insufficient as the paper describes the instances used for experiments but does not provide explicit details on training, validation, or test dataset splits, such as specific percentages, sample counts, or cross-validation setups.
Hardware Specification No The paper states, 'We implemented all domains and algorithms in Kotlin (Jet Brains 2017) and ran them on a modern Xeon server.' This description lacks specific hardware details such as the CPU model, number of cores, memory, or GPU specifications.
Software Dependencies Yes We implemented all domains and algorithms in Kotlin (Jet Brains 2017)
Experiment Setup Yes The parameter k was arbitrarily set to 10 for these experiments. and Each algorithm was given a maximum of 10 CPU minutes and 20 GB of RAM. and The stage expansion budget, initially set arbitrarily to ten nodes.