Expected Value of Communication for Planning in Ad Hoc Teamwork

Authors: William Macke, Reuth Mirsky, Peter Stone11290-11298

AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Lastly, this paper presents empirical results showing the performance of the new EDP-based algorithm in these complex settings, showing that it outperforms existing heuristics in terms of total number of steps required for the team to reach its goal.
Researcher Affiliation Collaboration William Macke,1 Reuth Mirsky, 1 Peter Stone 1,2 1 The University of Texas at Austin 2 Sony AI {wmacke,reuth,pstone}@cs.utexas.edu
Pseudocode Yes Algorithm 1 Query Policy
Open Source Code No The paper provides an arXiv link to the paper itself (ar Xiv:2103.01171) but does not provide concrete access to source code for the methodology described.
Open Datasets No The paper refers to a 'tool fetching domain' and 'domain instances' which are part of a simulated environment, rather than a publicly available dataset with specific access information.
Dataset Splits No The paper describes running experiments averaged over '100 domain instances' but does not specify dataset split information (percentages, sample counts, or detailed splitting methodology) for training, validation, or testing.
Hardware Specification Yes On average, all heuristic methods took < 0.23 seconds to complete each simulation, while e ZQ Query took on average 8.9 seconds on an Ubuntu 16.04 LTS Intel core i7 2.5 GHz
Software Dependencies No The paper mentions 'Ubuntu 16.04 LTS' as the operating system, but does not provide specific ancillary software details like library names with version numbers.
Experiment Setup Yes We ran experiments in a 20 20 grid, with 50 workstations and 5 toolboxes. Locations of the stations, toolboxes, and agents in each domain instance are chosen randomly. All results are averaged over 100 domain instances.