Agent-based Simulation Research Articles

Exploring the importance of stochasticity to Hybrid Equilibria in a Discrete Signaling Game

Abstract Communication via evolved signals is ubiquitous (both within and between species) in the natural world. However, how honest we should expect signals to be remains an open question. Hybrid equilibria are a form of equilibria predicted by discrete signaling games in which signalers are sometimes dishonest and signals do not completely reliably convey information on signaler quality. While these equilibria have been theoretically demonstrated in several signaling games, their dynamics in a stochastic simulation of evolutionary trajectories (that include representation of the inherent noise expected in evolution in the natural world) have not previously been studied. In this paper, we present an agent-based simulation of a discrete signaling game which exhibits hybrid equilibria. We find that while hybrid equilibria are evolutionarily attractive where they exist, populations exhibit variable and often drastic oscillating behavior around the predicted equilibrium values. We discuss how these dynamics might offer valuable opportunity for detecting hybrid equilibria in natural populations.

Sustainable Synchronization of Truck Arrival and Yard Crane Scheduling in Container Terminals: An Agent-Based Simulation of Centralized and Decentralized Approaches Considering Carbon Emissions

Background: Container terminal congestion is often measured by the average turnaround time for external trucks. Reducing the average turnaround time can be resolved by controlling the yard crane operation and the arrival times of external trucks (truck appointment system). Because the truck appointment system and yard crane scheduling problem are closely interconnected, this research investigates synchronization between the approaches used in truck appointment systems and yard crane scheduling strategies. Rubber-tired gantry (RTG) operators for yard crane scheduling operations strive to reduce RTG movement time as part of the container retrieval service. However, there is a conflict between individual agent goals. While seeking to minimize truck turnaround time, RTGs may travel long distances, ultimately slowing down the RTG service. Methods: We address a method that balances individual agent goals while also considering the collective objective, thereby minimizing turnaround time. An agent-based simulation is proposed to simulate scenarios for yard crane scheduling strategies and truck appointment system approaches, which are centralized and decentralized. This study explores the combined effects of different yard scheduling strategies and truck appointment procedures on performance indicators. Various configurations of the truck appointment system and yard scheduling strategies are modeled to investigate how those factors affect the average turnaround time, yard crane utilization, and CO2 emissions. Results: At all levels of truck arrival rates, the nearest-truck-first-served (NTFS) scenario tends to provide lower external truck turnaround times than the first-come-first-served (FCFS) and nearest-truck longest-waiting-time first-served (NLFS) scenario. Conclusions: The decentralized truck appointment system (DTAS) generally shows slightly higher efficiency in emission reduction compared with centralized truck appointment system (CTAS), especially at moderate to high truck arrival rates. The decentralized approach of the truck appointment system should be accompanied by the yard scheduling strategy to obtain better performance indicators.

Combining sterile insect releases with refuge areas to delay the evolution of resistance to Bt sugarcane: an agent-based modeling approach.

The strategic use of refuge areas is a well-known method for delaying the development of pest resistance to Bacillus thuringiensis (Bt) crop. A lesser-known method to control against resistance development is sterile insect releases. In this article, an agent-based simulation model is used to test the effectiveness of combining the use of Bt sugarcane, refuge areas, and sterile insect releases as an integrated strategy against Eldana saccharina Walker (Lepidoptera: Pyralidae) infestation and resistance development. Individual insects are modeled with their own genetic traits on a simulated sugarcane field that represents either Bt or refuge area. The model is applied to 2 hypothetical case studies. In the first experiment, resistance development and infestation dynamics in Bt sugarcane without refuge areas are considered using various sterile:wild sterile release ratios, and different release distributions. In the second experiment, the inclusion of a refuge area in Bt sugarcane is considered, using various sterile:wild releases ratios and different release distributions. A trade-off between sterile insect releases and the use of the refuge area was observed, and could, in some cases, work against each other.

Quantifying traffic emission reductions and traffic congestion alleviation from high-capacity ride-sharing

Despite the promising benefits that ride-sharing offers, there has been a lack of research on the benefits of high-capacity ride-sharing services. Prior research has also overlooked the relationship between traffic volume and the degree of traffic congestion and emissions. To address these gaps, this study develops an open-source agent-based simulation platform and a heuristic algorithm to quantify the benefits of high-capacity ride-sharing with significantly lower computational costs. The simulation platform integrates a traffic emission model and a speed-density traffic flow model to characterise the interactions between traffic congestion levels and emissions. The experiment results demonstrate that ride-sharing with vehicle capacities of 2, 4, and 6 passengers can alleviate total traffic congestion by approximately 3%, 4%, and 5%, and reduce traffic emissions of a ride-sourcing system by approximately 30%, 45%, and 50%, respectively. This study can guide transportation network companies in designing and managing more efficient and environment-friendly mobility systems.

On the impact of zealots in a population of susceptible agents in a best-of-n problem within a heterogeneous network

Both humans and social animals live in groups and are frequently forced to choose between options with different qualities. When there are no leader agents controlling the group decision, consensus can be achieved through repeated interactions among group members. Various studies on collective decision-making illustrate how the dynamics of the opinions are determined by the structure of the social network and the methods that individuals use to share and update their opinion upon a social interaction. In this paper, we are interested in further exploring how cognitive, social, and environmental factors interactively contribute to determining the outcome of a collective best-of-n decision process involving asymmetric options, i.e., different costs and/or benefits for each option. We propose and study a novel model capturing those different factors, i) the cognitive load in processing social information, ii) the number of zealots (i.e., asocial agents who never change their opinion), iii) the option qualities, iv) the social connectivity structure, and v) the degree centrality of the asocial agents (i.e., the number of neighbours). By using the heterogeneous mean-field approach, we study the impact of the above-mentioned factors in the decision dynamics. Our findings indicate that when susceptible agents, i.e., individuals who change their opinion to conform with others, use the voter model as a mechanism to update their opinion, both the number and the degree of connectivity of the zealots can lead the population to converge towards the lowest quality option. Instead, when susceptible agents use methods demanding a larger cognitive cost (e.g., the majority rule), the group is marginally impacted by the presence of zealots. The results of the analytical model are complemented and extended by agent-based simulations. Our analysis also shows that the network topology can modulate the influence of zealots on group dynamics. In fact, in homogeneous networks where all nodes have the same degree, any location of the zealots has similar impact on the group dynamics. Instead, when the network is heterogeneous, our simulations confirm the model predictions showing that placing the zealots in the network hubs (nodes with several neighbours) has a much larger impact than placing them in lower-degree nodes.

Exploring the Role of Genetic and Environmental Features in Colorectal Cancer Development: An Agent-Based Approach

The complexity of issues in cancer research has led to the introduction of powerful computational tools to help experimental in vivo and in vitro methods. These tools, which typically focus on studying cell behavior and dynamic cell populations, range from systems of differential equations that are solved numerically to lattice models and agent-based simulations. In particular, agent-based models (ABMs) are increasingly used due to their ability to incorporate multi-scale features, ranging from the individual to the population level. This approach allows for the combination of statistically aggregated assumptions with individual heterogeneity. In this work, we present an ABM that simulates tumor progression in a colonic crypt, to provide an experimental in silico environment for testing results achieved in traditional laboratory research and developing alternative scenarios of tumor development. The model also allows some speculations about causal relationships in biologically inspired systems.

From agent-based simulations of disease dynamics to expert support on public health interventions

Abstract Agent-based epidemiological simulators have long been used to predict the dynamics of infectious diseases. They allow to combine compartment approaches with representing contact networks derived from real or synthetic populations. Often, scenario techniques are applied to model the impact of public health interventions. During the COVID-19 pandemics the agent-based simulator COVASIM has been established for modelling the spreading of SARS-CoV-2. It includes four different contact networks related to homes, schools, workplaces and environment. During the pandemic it became obvious that the current simulation techniques were not sufficient to cover important needs in public health administrations and society. While most simulations were based on scenarios predicting expected numbers for the infected, hospitalized or critically ill, these studies could not provide easily strategic options on a variety of equivalent public health interventions, which are all suitable to keep disease dynamics within acceptable bounds, set e.g. by hospital capacity. Knowing about a variety of equivalent options, however, is of crucial importance to enable a fair burden sharing throughout all groups of society. Therefore, we applied machine learning techniques to create an inverse model for epidemiological simulations based on COVASIM. For given transmission rates it outputs a range of quantitative interventions which are equivalently suitable to keep epidemiological dynamics within predefined bounds. Currently, it covers interventions like school and workplace closings and other mobility restrictions. Future work will be directed to include other interventions, like vaccination campaigns, and we will work to make the approach more generalizable for other infectious diseases. The tool may turn out helpful fur public health administrations and stimulate the public debate on the best way a society may take in a pandemic event or other situations where public health action has to be imposed. Key messages • Machine Learning allows to invert agent based simulators. • Instead of disease dynamics simulations a set of equivalent public health interventions consistent with pre-defined constraints can be calculated.

Understanding the evolution of engagement with brand posts on social media using agent-based models

Purpose This paper aims to delineate the nature of the costs associated with the diffusion of brand posts. Since cost is an essential component of customer lifetime value, this research also aims to identify factors that impact consumer engagement (CE) with brand posts. Design/methodology/approach This study conceptualize marginal cost of CE as the ratio of a unit change in cost for a unit change in CE – this ratio is conceptualized for a paid brand post compared to an organic brand post. The authors present a theoretically grounded simulation method that analyzes the relative impact of several factors that influence CE with brand posts. By using an agent-based simulation model, the method presented in this paper helps explain the relative impact of four factors on CE with brand posts on social media. Findings The authors explain four factors that influence the diffusion of CE with a brand post – promotion-related (i.e. whether it is paid vs organic), network-related, user-related and active periods. The authors also show that the effectiveness of paid posts vs organic posts attributed to a better audience assembled by the seeding algorithms dissipates after the initial stages of diffusion. These findings indicate that paid posts are effective vs organic posts in large part due to higher exposure than due to the efficiency of the algorithms to assemble nodes that are likely to engage with the post. Research limitations/implications An agent-based model details the impact of several factors that influence the effectiveness of paid and organic posts and presents methods to analyze the impact of these factors that have implications for theory. Practical implications The authors address the difficulty in ascertaining the effectiveness of paid posts by proposing a metric – marginal cost of engagement, defined as the incremental cost incurred by the marketer in earning one additional unit of CE. Since the costs incurred on paid and organic posts differ only on variable components, the marginal cost of engagement presents a way to benchmark paid posts against organic posts, which is very useful for managers in ascertaining the amount to spend on paid posts on social media. Originality/value There is little prior research on the costs involved in the diffusion of brand posts on social media. Identifying specific factors that impact costs is valuable for theory and practice.

AI-driven adaptive analysis for finding emergent behavior in military capability design

This article introduces and demonstrates a new methodology for searching for emergent behavior in simulation-based analysis as rare, extreme events using adaptive techniques enabled by recent advances in Bayesian machine learning (ML). The new methodology, Low-cost Adaptive exploratioN to Track down Extreme, Rare events using Numerical optimization (LANTERN), supports analysis activities in defense planning that iteratively build up the understanding of new technology and concept alternatives in complex military scenarios. Central to this process is emergent behavior—hard-to-predict but highly important behaviors that present problems or opportunities. These unexpected behaviors can be generated in complex military scenarios and are crucial to decision-making, but are often difficult to find and work with due to the expense and cost of the existing approaches for working with high-fidelity military simulation. To address this challenge, LANTERN is formulated to accelerate the discovery of emergent behavior as rare, extreme events by combining human expert understanding with new artificial intelligence (AI)-driven adaptive experimentation techniques in iterative analysis. A demonstration of the methodology is presented using military agent-based simulation scenarios developed in the Advanced Framework for Simulation, Integration, and Modeling (AFSIM). The demonstration highlights how analysis can focus directly on searching for emergent behavior and shows substantial improvements over brute-force Monte Carlo approaches.

Agent-Based Geosimulation of Yayoi Period Population Dynamics in Kyushu, Japan

Current scholarship in Japanese archaeology recognizes the people and culture of the Yayoi period as being descended from both continental migrants (Toraijin) and the native inhabitants (Jomon people). However, the number of migrants that came to Japan and the degree to which they were directly responsible for the development of Yayoi culture is a topic of debate. Using northern Kyushu as a case study, the author employs an agent-based simulation based on geospatial data to model the spread of population, rice agriculture, and continental genetics during the Early Yayoi period. Multiple simulations are performed with varying parameters related to migrant population size, marriage customs, and the ratio of men to women amongst migrants to test their effect on the demographics of the Yayoi period. Simulation results were compared to population estimates and the distribution of mitochondrial DNA in order to determine their validity. Experiment 1 showed that the average number of continental migrants was likely in the order of magnitude of 10s of people per year (approx. 40 to 50) during the first 500 years of the Yayoi period. Experiment 2 demonstrated that polygyny can lead to increased population growth and aids in the spread of continental genetics, but without more rigorous evaluation criteria it is difficult to conclude whether it was practiced during the period. The methods used in this study demonstrate how agent-based simulations can model realistically sized populations moving through geographic space by basing simulation inputs on geospatial data and population estimates.

Coordination mechanisms applied to logistical systems for local disaster preparedness: a Latin American case

Purpose The study aims to present an agent-based simulation model (ABM) for exploring interorganizational coordination scenarios in local disaster preparedness. This approach includes local actors and logistical processes as agents to compare various strategic coordination mechanisms. Design/methodology/approach The ABM model, developed in the Latin American context, specifically focuses on a case study of Colombia. Three coordination mechanisms (centralized, decentralized and cluster-type) have been evaluated using three performance indicators: effectiveness, efficiency and flexibility. Findings Simulation results show that the decentralized scenario outperforms in terms of efficiency and flexibility. On the contrary, the centralized and cluster-type scenarios demonstrate higher effectiveness, achieving a greater percentage of requirements coverage during the disaster preparedness stage. The ABM approach effectively evaluates strategical coordination mechanisms based on the analyzed performance indicators. Research limitations/implications This study has limitations due to the application of results to a single real case. In addition, the focus of the study is primarily on a specific type of disaster, specifically hydrometeorological events such as flash floods, torrential rains and landslides. Moreover, the scope of decision-making is restricted to key actors involved in local-level disaster management within a municipality. Originality/value The proposed ABM model has the potential as a decision-making tool for policies and local coordination schemes for future disasters. The simulation tool could also explore diverse geographical scenarios and disaster types, demonstrating its versatility and broader applicability for further insights and recommendations.

EOS: Impact Evaluation of Electric Vehicle Adoption on Peak Load Shaving Using Agent-Based Modeling

The increasing adoption of electric vehicles (EVs) by the general population creates an opportunity to deploy the energy storage capability of EVs for performing peak energy shaving in their households and ultimately in their neighborhood grid during surging demand. However, the impact of the adoption rate in a neighborhood might be counterbalanced by the energy demand of EVs during off-peak hours. Therefore, achieving optimal peak energy shaving is a product of a sensitive balancing process that depends on the EV adoption rate. In this paper, we propose EOS, an agent-based simulation model, to represent independent household energy usage and estimate the real-time neighborhood energy consumption and peak shaving energy amount of a neighborhood. This study uses Residential Energy Consumption Survey (RECS) and the American Time Use Survey (ATUS) data to model realistic real-time household energy use. We evaluate the impact of the EV adoption rates of a neighborhood on performing energy peak shaving during sudden energy surges. Our findings reveal these trade-offs and, specifically, a reduction of up to 30% of the peak neighborhood energy usage for the optimal neighborhood EV adoption rate in a 1089 household neighborhood.

Estimation of Domestic Load Profile for Effective Demand-Side Management

In this article domestic base and shift based load profile are estimated for effective Demand Side Management (DSM). Domestic electricity consumption in Pakistan accounts for approximately 40% of total electricity sales. Effective DSM can reduce electricity consumption. Power generation companies are increasingly recognising the importance of analysing their customers' consumption patterns. This article proposes an agent-based modelling approach for forecasting household electricity consumption profile. AnyLogic software is used for establishing the model that uses variables and functions to estimate base and shiftable load of three hundred households. This article also estimates the weekly and 24- hour load profile for effective DSM operations by smart grid infrastructure. The results show that the agent-based simulation closely approximates real-time data, with a difference of only 2.4%.

Mapping secondary data gaps for social simulation modelling: A case study of the journeys of Syrian asylum seekers to Europe

Simulation models of social processes may require data that are not readily available, have low accuracy, are incomplete or biased. The paper presents a formal process for collating, assessing, selecting, and using secondary data as part of creating, validating, and documenting an agent-based simulation model of a complex social process, in this case, asylum seekers’ journeys to Europe. The process starts by creating an inventory of data sources, and the associated metadata, followed by assessing different aspects of data quality according to pre-defined criteria. As a result, based on the typology of available data, we are able to produce a thematic map of the area under study, and assess the uncertainty of key data sources, at least qualitatively. We illustrate the process by looking at the data on Syrian migration to Europe in 2011–21. In parallel, successive stages of the development of a simulation model allow for identifying key types of information which are needed as input into empirically grounded modelling analysis. Juxtaposing the available evidence and model requirements allows for identifying knowledge gaps that need filling, preferably by collecting additional primary data, or, failing that, by carrying out a sensitivity analysis for the assumptions made. By doing so, we offer a way of formalising the data collection process in the context of model-building endeavours, while allowing the modelling to be predominantly question-driven rather than purely data-driven. The paper concludes with recommendations with respect to data and evidence, both for modellers, as well as model users in practice-oriented applications.

GenSynthPop: generating a spatially explicit synthetic population of individuals and households from aggregated data

Synthetic populations are representations of actual individuals living in a specific area. They play an increasingly important role in studying and modeling individuals and are often used to build agent-based social simulations. Traditional approaches for synthesizing populations use a detailed sample of the population (which may not be available) or combine data into a single joint distribution, and draw individuals or households from these. The latter group of existing sample-free methods fail to integrate (1) the best available data on spatial granular distributions, (2) multi-variable joint distributions, and (3) household level distributions. In this paper, we propose a sample-free approach where synthetic individuals and households directly represent the estimated joint distribution to which attributes are iteratively added, conditioned on previous attributes such that the relative frequencies within each joint group of attributes are maintained and fit granular spatial marginal distributions. In this paper we present our method and test it for the Zuid-West district of The Hague, the Netherlands, showing that spatial, multi-variable and household distributions are accurately reflected in the resulting synthetic population.

Implication of makerspace layout design in rural primary schools of Lianyungang, Jiangsu Province, China

ABSTRACT Makerspaces originated from students’ interests and utilised digital tools through project-based learning, advocating creation, encouraging sharing, and cultivating students’ interdisciplinary problem-solving, teamwork, and innovation skills. Makerspaces aligned with China's educational reforms to support different teaching and learning (TnL) activities. This study examines the impact of different spatial layouts, furniture, and technology arrangements in makerspaces on TnL activities in rural primary schools in Lianyungang, Jiangsu Province, China. By analysing four TnL activities, which are Presentations, Discussions, Individual Learning, and Team-Teaching, this research seeks to improve educational performances in remote areas. Using quantitative data and AnyLogic agent-based simulations, this study assesses the effectiveness of different makerspace models in three case study schools in rural area Lianyungang. The findings reveal that the effectiveness of makerspace models varies significantly in improving student engagement and performance, particularly in Presentation and Team-Teaching activities. Results suggest that different makerspace models can vary significantly in their effectiveness in improving student engagement and performance, especially in Presentation and Team-Teaching activities. The study offers insights on how to structure these spaces for twenty-first-century learning and highlights adaptable makerspaces that meet the evolving needs of both students and teachers. This research findings are crucial for optimising the makerspaces design and adjusting educational practices in the context of Chinese educational reform. The results also have far-reaching implications in informing educational management and policymaking, providing practical solutions for designing makerspaces that address the needs of teachers and students in rural areas. This research provides valuable recommendations for improving rural education by implementing makerspace.

Assessing port cluster resilience: Integrating hypergraph-based modeling and agent-based simulation

With increasing global trade and frequent occurrence of disruptive events, the resilience of port clusters has emerged as a critical area of concern. However, studies that focus on the resilience of port clusters considering their complex network structure and operational dynamics remain limited. This study proposes a novel model to assess port cluster resilience by integrating hypergraph-based modeling and agent-based simulation. The model captures the complex relationships among ports and vessels, enabling the dynamic modeling of disruption impacts on port cluster resilience. A case study of the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) port cluster demonstrates the model’s applicability and effectiveness. Additionally, the significant impact of typhoon duration on resilience and the potential benefits of vessel port skipping behavior and port cargo handling capacity improvements are analyzed. These findings provide valuable insights for stakeholders in developing effective strategies to enhance the resilience of port clusters and the maritime transportation system.

Application of simulation and machine learning in supply chain management: A synthesis of the literature using the Sim-ML literature classification framework

Stochastic modeling techniques, such as discrete-event and agent-based simulation, are widely used in supply chain management (SCM) for capturing real-world uncertainties. Over the last decade, data-driven approaches like machine learning (ML) have also gained prominence in SCM, employing methods such as supervised learning (SL), unsupervised learning (UL), and reinforcement learning (RL). As supply chains grow in complexity, hybrid models combining simulation (Sim) and ML are becoming increasingly common, and the field stands to gain from a structured review of this literature. Towards this, we developed the Sim-ML Literature Classification Framework, which includes a hierarchical taxonomy comprising five SC criteria, 22 Sim-ML classes and over 75 Sim-ML subclasses. We applied this framework to synthesize 99 papers, revealing significant diversity in how Sim-ML models are used to address supply chain challenges. Key findings include the recognition of the breadth of study objectives, identifying various forms of model hybridization achieved by combining discrete/continuous simulation techniques with SL, UL, and RL approaches, and the data flow mechanisms such as sequential and feedback methods employed by the simulation and ML elements of the hybrid model. Our findings also identify some gaps in the literature; for example, optimization is rarely incorporated into Sim-ML models. Also, most studies present Sim-ML models for addressing problems in general supply chains, likely due to the lack of access to industrial data. The review also highlights that Industry 4.0 technologies, such as digital twins and blockchain, are underrepresented in current research, as are topics like sustainability and transportation. These gaps suggest significant opportunities for future research. We provide guidelines for practitioners on applying Sim-ML models to manage supply chain drivers, mitigate the impact of disruptions, and integrate emerging technologies. Our review serves as a valuable resource for researchers, practitioners, and students interested in leveraging Sim-ML approaches in SCM.

Exploring Trade-Offs for Online Mental Health Matching: Agent-Based Modeling Study.

Online mental health communities (OMHCs) are an effective and accessible channel to give and receive social support for individuals with mental and emotional issues. However, a key challenge on these platforms is finding suitable partners to interact with given that mechanisms to match users are currently underdeveloped or highly naive. In this study, we collaborated with one of the world's largest OMHCs; our contribution is to show the application of agent-based modeling for the design of online community matching algorithms. We developed an agent-based simulation framework and showcased how it can uncover trade-offs in different matching algorithms between people seeking support and volunteer counselors. We used a comprehensive data set spanning January 2020 to April 2022 to create a simulation framework based on agent-based modeling that replicates the current matching mechanisms of our research site. After validating the accuracy of this simulated replication, we used this simulation framework as a "sandbox" to test different matching algorithms based on the deferred acceptance algorithm. We compared trade-offs among these different matching algorithms based on various metrics of interest, such as chat ratings and matching success rates. Our study suggests that various tensions emerge through different algorithmic choices for these communities. For example, our simulation uncovered that increased waiting time for support seekers was an inherent consequence on these sites when intelligent matching was used to find more suitable matches. Our simulation also verified some intuitive effects, such as that the greatest number of support seeker-counselor matches occurred using a "first come, first served" protocol, whereas relatively fewer matches occurred using a "last come, first served" protocol. We also discuss practical findings regarding matching for vulnerable versus overall populations. Results by demographic group revealed disparities-underaged and gender minority groups had lower average chat ratings and higher blocking rates on the site when compared to their majority counterparts, indicating the potential benefits of algorithmically matching them. We found that some protocols, such as a "filter"-based approach that matched vulnerable support seekers only with a counselor of their same demographic, led to improvements for these groups but resulted in lower satisfaction (-12%) among the overall population. However, this trade-off between minority and majority groups was not observed when using "topic" as a matching criterion. Topic-based matching actually outperformed the filter-based protocol among underaged people and led to significant improvements over the status quo among all minority and majority groups-specifically, a 6% average chat rating improvement and a decrease in blocking incidents from 5.86% to 4.26%. Agent-based modeling can reveal significant design considerations in the OMHC context, including trade-offs in various outcome metrics and the potential benefits of algorithmic matching for marginalized communities.

Demand-Aware Distributed Pathfinding for Repositioning Vehicles in Shared-use Autonomous Mobility Services

Shared-use autonomous mobility services (SAMS) have the potential to provide accessible and demand-responsive mobility to passengers, while benefitting from autonomous vehicle (AV) technology and bypassing challenges related to supply-side incentives or individual driver goals. SAMS operators typically aim to achieve efficiency and improved service quality in their fleet operations, both of which are further enabled by the use of AVs. Specifically, fleet repositioning decisions in anticipation of future demand can improve service quality, but existing approaches in the literature seldom consider the problem of routing repositioning vehicles in a way that further improves SAMS objectives. This paper presents an approach for demand-aware distributed routing pathfinding for repositioning vehicles, which can supplement existing vehicle repositioning approaches. The problem is formulated with a multi-criteria objective that minimizes the vehicles’ total travel time and maximizes their total demand-serving potential, while distributing that potential equitably among the ride-seeking passengers across the transportation network. We evaluate the proposed approach via numerical experiments using an agent-based simulation of SAMS operations in the network of Manhattan in New York City. The proposed approach is compared to a baseline simple shortest path approach for routing the repositioning vehicles. The results demonstrate that mean passenger waiting times for pick-up can be reduced, while also reducing the total vehicle miles and the empty miles travelled due to repositioning. Thus, the proposed approach can help improve the overall system performance in terms of both service quality and efficiency metrics, relative to the baseline approach.