Collaborative Planning with Encoding of Users' High-level Strategies
J. Kim, C. Banks, and J. Shah. Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence and the Twenty-Ninth Innovative Applications of Artificial Intelligence Conference, San Francisco, California, USA, page 955-961. Palo Alto, California, AAAI Press, (2017)
Abstract
The generation of near-optimal plans for multi-agent systems with numerical states and temporal actions is computationally challenging. Current off-the-shelf planners can take a very long time before generating a near-optimal solution. In an effort to reduce plan computation time, increase the quality of the resulting plans, and make them more interpretable by humans, we explore collaborative planning techniques that actively involve human users in plan generation. Specifically, we explore a framework in which users provide high-level strategies encoded as soft preferences to guide the low-level search of the planner. Through human subject experimentation, we empirically demonstrate that this approach results in statistically significant improvements to plan quality, without substantially increasing computation time. We also show that the resulting plans achieve greater similarity to those generated by humans with regard to the produced sequences of actions, as compared to plans that do not incorporate user-provided strategies.
Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence and the Twenty-Ninth Innovative Applications of Artificial Intelligence Conference, San Francisco, California, USA
year
2017
pages
955-961
publisher
AAAI Press
xeditor
Singh, Satinder and Markovitch, Shaul
file
Preprint/AAAI Digital Library:2017/KimBanksShah17AAAI.pdf:PDF
%0 Conference Paper
%1 KimBanksShah17AAAI
%A Kim, Joseph
%A Banks, Christopher J.
%A Shah, Julie A.
%B Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence and the Twenty-Ninth Innovative Applications of Artificial Intelligence Conference, San Francisco, California, USA
%C Palo Alto, California
%D 2017
%I AAAI Press
%K 01801 aaai paper ai temporal plan algorithm user interaction
%P 955-961
%T Collaborative Planning with Encoding of Users' High-level Strategies
%U http://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14840
%X The generation of near-optimal plans for multi-agent systems with numerical states and temporal actions is computationally challenging. Current off-the-shelf planners can take a very long time before generating a near-optimal solution. In an effort to reduce plan computation time, increase the quality of the resulting plans, and make them more interpretable by humans, we explore collaborative planning techniques that actively involve human users in plan generation. Specifically, we explore a framework in which users provide high-level strategies encoded as soft preferences to guide the low-level search of the planner. Through human subject experimentation, we empirically demonstrate that this approach results in statistically significant improvements to plan quality, without substantially increasing computation time. We also show that the resulting plans achieve greater similarity to those generated by humans with regard to the produced sequences of actions, as compared to plans that do not incorporate user-provided strategies.
@inproceedings{KimBanksShah17AAAI,
abstract = {The generation of near-optimal plans for multi-agent systems with numerical states and temporal actions is computationally challenging. Current off-the-shelf planners can take a very long time before generating a near-optimal solution. In an effort to reduce plan computation time, increase the quality of the resulting plans, and make them more interpretable by humans, we explore collaborative planning techniques that actively involve human users in plan generation. Specifically, we explore a framework in which users provide high-level strategies encoded as soft preferences to guide the low-level search of the planner. Through human subject experimentation, we empirically demonstrate that this approach results in statistically significant improvements to plan quality, without substantially increasing computation time. We also show that the resulting plans achieve greater similarity to those generated by humans with regard to the produced sequences of actions, as compared to plans that do not incorporate user-provided strategies.},
added-at = {2018-02-23T13:42:48.000+0100},
address = {Palo Alto, California},
author = {Kim, Joseph and Banks, Christopher J. and Shah, Julie A.},
biburl = {https://www.bibsonomy.org/bibtex/2e9642182305ba2fe8e86e0b410109deb/flint63},
booktitle = {Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence and the Twenty-Ninth Innovative Applications of Artificial Intelligence Conference, San Francisco, California, USA},
file = {Preprint/AAAI Digital Library:2017/KimBanksShah17AAAI.pdf:PDF},
groups = {public},
interhash = {ca96e94e281ee14bc812663f63444d46},
intrahash = {e9642182305ba2fe8e86e0b410109deb},
keywords = {01801 aaai paper ai temporal plan algorithm user interaction},
pages = {955-961},
publisher = {AAAI Press},
timestamp = {2018-04-16T11:45:03.000+0200},
title = {Collaborative Planning with Encoding of Users' High-level Strategies},
url = {http://aaai.org/ocs/index.php/AAAI/AAAI17/paper/view/14840},
username = {flint63},
xeditor = {Singh, Satinder and Markovitch, Shaul},
year = 2017
}