TY - GEN
T1 - Experimental Validation of a Multi-objective Planning Decision Support System for Ship Routing Under Time Stress
AU - Macesker, Matthew
AU - Pattipati, Krishna R.
AU - Sidoti, David
AU - Bienkowski, Adam
AU - Zhang, Lingyi
AU - Kleinman, David L.
AU - McGuire, Mollie
AU - Uziel, Steven
AU - Basu Roy, Senjuti
AU - Primerano, Francesco
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - Integration of sophisticated planning algorithms into Naval operations requires the systematic design of decision-support systems (DSS) that improve the understanding of AI-suggested courses of action (CoAs) by humans without overwhelming their cognitive processes. A successful interface would permit operators to fully understand the context of the problem so that they can select a computer-generated CoA that best matches their preferences. In this paper, the authors investigated such a system for routing ships using two sequential human-in-the-loop experiments, one which evaluated the impact of various forms of graphical decision support on decision-making and the cognitive load, and another in which time pressure was manipulated. The results showed that a mix between tabular and graphical information reduced the cognitive load, given adequate time to make a decision. Participant responses were used to build models of human decision rules to integrate into the DSS, revealing that humans heavily weighted certain contextual attributes that were indirectly integrated into the planning algorithm through the cost structure. A novel technique for representing decisions as a distribution of common heuristic trade-offs among Pareto solutions found that the context of each scenario dictated the choice of the heuristic. The results of these experiments guided the design of a follow-up experiment on multi-ship routing that is currently in pilot testing.
AB - Integration of sophisticated planning algorithms into Naval operations requires the systematic design of decision-support systems (DSS) that improve the understanding of AI-suggested courses of action (CoAs) by humans without overwhelming their cognitive processes. A successful interface would permit operators to fully understand the context of the problem so that they can select a computer-generated CoA that best matches their preferences. In this paper, the authors investigated such a system for routing ships using two sequential human-in-the-loop experiments, one which evaluated the impact of various forms of graphical decision support on decision-making and the cognitive load, and another in which time pressure was manipulated. The results showed that a mix between tabular and graphical information reduced the cognitive load, given adequate time to make a decision. Participant responses were used to build models of human decision rules to integrate into the DSS, revealing that humans heavily weighted certain contextual attributes that were indirectly integrated into the planning algorithm through the cost structure. A novel technique for representing decisions as a distribution of common heuristic trade-offs among Pareto solutions found that the context of each scenario dictated the choice of the heuristic. The results of these experiments guided the design of a follow-up experiment on multi-ship routing that is currently in pilot testing.
KW - Association Rule Mining
KW - Collaborative AI
KW - Decision Support Systems
KW - Human-in-the-loop Experiment
KW - Pareto-optimal
KW - Ship Routing
KW - TMPLAR
UR - http://www.scopus.com/inward/record.url?scp=85173011737&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85173011737&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-35894-4_26
DO - 10.1007/978-3-031-35894-4_26
M3 - Conference contribution
AN - SCOPUS:85173011737
SN - 9783031358937
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 346
EP - 365
BT - Artificial Intelligence in HCI - 4th International Conference, AI-HCI 2023, Held as Part of the 25th HCI International Conference, HCII 2023, Proceedings
A2 - Degen, Helmut
A2 - Ntoa, Stavroula
PB - Springer Science and Business Media Deutschland GmbH
T2 - 4th International Conference on Artificial Intelligence in HCI, AI-HCI 2023, held as part of the 25th International Conference on Human-Computer Interaction, HCII 2023
Y2 - 23 July 2023 through 28 July 2023
ER -