ABSTRACT. In collective problem-solving, agents engage in a sequential search process that combines individual exploration with group knowledge utilization to generate innovative solutions. The equilibrium between independent exploration and reliance on the group (exploitation) largely depends on multiple factors, chiefly the complexity of the problem being tackled. Existing research suggests that, as the complexity of the problem increases, there is an advantage in agents favoring individual exploration. However, the extent to which agents behaviorally reduce their dependence on the group's collective intelligence in highly complex scenarios remains uncertain. Our hypothesis posits that, as complexity escalates, agents may lean more toward minimizing risk and cognitive effort by exploiting group knowledge rather than embracing the uncertainty of individual exploration. This essentially mirrors a social dilemma akin to the classic public goods game, where individual incentives clash with the collective good. To test this hypothesis, we conducted a pre-registered human-subject experiment comprising a two-stage lab-based study. In the first stage, innovators, our participants, assessed the complexity of a given task, and in the second stage, they indicated their propensity for group collaboration and reliance on collective intelligence during decision-making. Our findings reveal that when participants face weakly dominant incentives, they tend to choose exploitation over exploration, which is particularly pronounced in complex scenarios and when personal search performance expectations are moderate. This behavior holds significant policy implications, particularly in the context of designing engineering-teams seeking to harness the benefits of collective intelligence; so, we examine the conditions under which group-reliance behavior intensifies with complexity.

ABSTRACT. Decision-based design typically views decision problems under uncertainty through the lens of traditional normative decision theory. This typically entails the use of the mean of a value or utility distribution as the decision criterion. Axiomatic arguments can be made in favor of such mean-based schemes, as well as, arguments invoking the central limit theorem. Where these arguments can encounter limitations, however, are decisions with a finite number of repetitions. This presentation illustrates this limitation through two examples: a classic St. Petersburg game and a component reliability example. Each example will be evaluated with different amounts of repetition, and the results are analyzed to show how the representativeness of a distribution mean can be limited for low repetition problems with skewed outcome distributions. Even when a risk-averse utility function is used to calculate an entrance fee, such a player would still be more likely to lose money in a St. Petersburg game than to gain money unless the game is repeated hundreds of times in a row. These results highlight the importance of understanding the assumptions made by decision making techniques as well as how well the specific conditions of an engineering problem meet those assumptions.

ABSTRACT. Quick and areawide understanding of post-disaster situation is one of the key factors for optimal allocation of rescue resources. Information is primarily obtained through on-site inspections by local government officials. However, this practice not only brings danger to the inspectors who need to be on disaster-stricken area, but also consumes time before information is transferred to decision makers.

To overcome these problems, this research proposes the Driving and Disaster (D2) Recorder system that utilizes pre-installed, vehicle-mounted driving recorders for disaster situation monitoring. The originality is the functionality to automatically focus and select only important information for decision makers, which is realized through image recognition and actuation.

Initial prototype test using a car-mounted driving recorder identified challenges from technical, legal and commercial perspectives. From technical aspect, the method to automatically evaluate data validity needs to be developed because the research on evaluation criteria and its enabler is underexplored. On legal aspect, the D2 Recorder system needs to be operated in secure and consensus-based manner because a part of techniques overlaps which that of illegal remote hacking. Thus, careful consensus building will be required when it comes to implementing this in society. On commercial aspect, installing additional functions could raise prices which could drop sales. Government subsidies could be a solution, however doing so also requires consensus with consumers.

Future research should address identified challenges while seeking possibility to cooperate with autonomous vehicles that could deliver on-site information from anywhere you need.

ABSTRACT. The intricacies of collective decision-making in systems-of-systems design pose challenges stemming from localized incentives, incomplete information, and decentralized control. Traditional optimization methods offer limited effectiveness in addressing these complexities, resulting in strategic uncertainty that impacts system performance and decision-making. This study identifies a critical research gap in harnessing the potential of game theory to illuminate and navigate these complexities. To bridge this gap, we extend the concept of structural fear and greed strategy dynamics from two-player games to multi-player scenarios. We introduce the concept of "strategic hindrance," a method to facilitate the evaluation of stability in collective action by analyzing players' beliefs about the game and other players' strategies, particularly in contexts where exhaustive knowledge about the links between players is elusive. We quantify structural fear and greed by dissecting normal-form games into player-reduced games, visualizing strategic hindrance spaces that offer insights into strategic trade-offs amid ambiguity and uncertainty. We apply this framework to a graphical volunteer's dilemma, demonstrating the evolution of unfavorable strategy dynamics with increasing player numbers, paralleling real-world phenomena like the bystander effect. We further show that we can induce favorable strategy dynamics and reduce the prevalence of undesirable equilibrium conditions by judiciously altering player connectivity. The findings of this research can play a vital role in enhancing our understanding of strategy dynamics in complex decision-making processes. The insights gained from this study can help design effective incentive mechanisms within system-of-systems engineering, which can lead to socially efficient outcomes.

ABSTRACT. Resilience is the ability of a system to respond, absorb, adapt, and recover from a disruptive event. This research presents three alternative approaches to model and predict performance, independent of the domain of application, and resilience metrics with techniques from reliability engineering, including (i) bathtub-shaped hazard functions, (ii) mixture distributions, and (iii) a model incorporating covariates related to the intensity of events that degrade performance as well as efforts to restore performance. A historical data set on job losses during the latest recession in the United States is used to assess the predictive accuracy of these approaches. Goodness-of-fit measures and confidence intervals as well as interval-based resilience metrics are computed to assess how well the models perform on the data sets considered. Results suggest that covariate models perform best, tracking trends most closely and predicting future performance most accurately.

ABSTRACT. Information management and diffusion are increasingly important focus areas for humanitarian response agencies and researchers. To assist in understanding the tradeoffs inherent to various humanitarian information management strategies, this research investigates how several strategies impact the speed of team data collection, and how organizational design affects the performance of these strategies. An agent-based model explores how each of these strategies contributes to faster task accomplishment and lower information overload. Findings suggest that holding meetings -- whether all-hands or in siloed teams -- was more important in speeding data collection than increasing willingness to exchange information with other organizations. In addition, introducing a dedicated information manager improved task completion time as effectively as meetings while also reducing the amount of unnecessary information each participant collected.

ABSTRACT. The high attrition rates at college-level science, technology, engineering, and mathematics (STEM) programs could undermine the U.S.’s economic and defense capabilities. In this work, we propose and demonstrate a system dynamics approach to simulate and test the impact of potential policy interventions to raise STEM retention. The model incorporates multiple data sources, studies, and regression analyses. This model was used to evaluate the impact (including uncertainty ranges) of potential policy interventions (scholarships, workshops, internships) on STEM degree attainment. The outcomes of this study highlight the capacity of our system dynamics model to provide valuable insights into the consequences of policy interventions and support decision-making.

ABSTRACT. Fleet electrification is urgently needed to achieve net-zero emissions and enhance sustainability, particularly in maritime container ports. These ports are actively exploring the integration of electric terminal tractors and associated infrastructure to minimize environmental impact and enhance service performance. However, challenges arise in planning infrastructure investments that can meet the charging needs of these tractors while preserving operational efficiency.

This paper presents an optimization framework and associated risk assessment for strategically expanding electric vehicle fleets at maritime container ports. The methodology incorporates multi-criteria decision analysis (MCDA) and characterizes enterprise risk as a disruption of system order. The optimization process utilizes linear programming models to evaluate thirty-two combinations of plug-in, wireless, and wireless dynamic charging infrastructure configurations to identify optimal charger locations.

A robust ensemble model accompanies the optimization, featuring a comprehensive risk analysis across seven scenarios, including Environmental Change, Policy Revision, Technology Innovation, Cyber Attack, Market Shift, Electrical Grid Stress, and Workforce Interruption. The results provide valuable insights to support decision-making and enterprise risk management for a $1.5 billion strategic port infrastructure plan. This plan entails selecting charging station locations, establishing charging schedules, and choosing charger models while considering key performance criteria like safety, operational efficiency, cost-effectiveness, and reliability.

This approach holds broad applicability for various complex systems, enabling the mitigation of schedule and cost risks while promoting sustainability. The paper's target audience includes transportation and energy infrastructure owners, operators, asset managers, logistics service providers, and other stakeholders.

ABSTRACT. Within the transportation sector, vehicle electrification is widely cited as a primary mitigation strategy for realizing greenhouse gas (GHG) reductions. Planning for the impact of vehicle charging on the electricity grid requires temporarily resolved charging demand profiles that reflect the travel decisions and charging behaviours of plug-in electric vehicle (PEV) owners. The simulation model developed in this research for New York State and New England integrates the travel and electricity systems by generating time-specific vehicle-charging profiles from retabulated household travel survey data. Electricity and travel demand are calculated for hourly time steps to match the temporal resolution of the regional electricity generating dispatch model. Scenarios modelled include different PEV adoption levels, charging preferences, and different charging station availability. Results indicate that when charging infrastructure was available at more locations, a larger portion of charging demand was shifted off-peak and into the morning hours relative to home-only charging scenarios. This was true even without incentives or managed pricing schemes. The results also provide evidence of the particular importance of workplace charging. Even in the scenario with universally available charging infrastructure, 39% of all non-home charging demand occurred at workplaces, and work stops had the highest percentage of non-home charging events. All scenarios result in increased peak demand and increased generation by non-renewable generating sources. This indicates that pricing or other incentive mechanisms that influence charging decisions could result in lower cost, lower emissions outcomes.