Multi-agent Transport Research Articles

Optimal roadworks schedule in multi-agent transportation models

In this paper we consider the problem of the optimal roadworks scheduling in the road traffic network optimization. The main goal of this research is to find the optimal order of roadworks and minimize their negative impact on the transportation network throughput. In order to model the dynamics of a transportation system, we propose a large-scale, multi-agent vehicle routing modelling framework for traffic simulation and impact of roadworks on traffic throughput. We present a quick and effective heuristic algorithm for streamlining the road repairs. Finally, the numeric experiment in a real-world setting is conducted. The obtained results indicate the flexibility of the proposed algorithm, which can be effectively applied by decision makers who need to carry out roadworks with limited time and resources.

Dynamic Calibration of a Carsharing System in Multiagent Transport Simulation

Dynamic Calibration of a Carsharing System in Multiagent Transport Simulation

RAMPART: Reinforcing Autonomous Multi-agent Protection through Adversarial Resistance in Transportation

In the field of multi-agent autonomous transportation, such as automated payload delivery or highway on-ramp merging, agents routinely exchange knowledge to optimize their shared objective and adapt to environmental novelties through Cooperative Multi-Agent Reinforcement Learning (CMARL) algorithms. This knowledge exchange between agents allows these systems to operate efficiently and adapt to dynamic environments. However, this cooperative learning process is susceptible to adversarial poisoning attacks, as highlighted by contemporary research. Particularly, the poisoning attacks where malicious agents inject deceptive information camouflaged within the differential noise, a pivotal element for differential privacy (DP)-based CMARL algorithms, pose formidable challenges to identify and overcome. The consequences of not addressing this issue are far-reaching, potentially jeopardizing safety-critical operations and the integrity of data privacy in these applications. Existing research has strived to develop anomaly detection-based defense models to counteract conventional poisoning methods. Nonetheless, the recurring necessity for model offloading and retraining with labeled anomalous data undermines their practicality, considering the inherently dynamic nature of the safety-critical autonomous transportation applications. Further, it is imperative to maintain data privacy, ensure high performance, and adapt to environmental changes. Motivated by these challenges, this paper introduces a novel defense mechanism against stealthy adversarial poisoning attacks in the autonomous transportation domain, termed Reinforcing Autonomous Multi-agent Protection through Adversarial Resistance in Transportation (RAMPART). Leveraging a GAN model at each local node, RAMPART effectively filters out malicious advice in an unsupervised manner, whilst generating synthetic samples for each state-action pair to accommodate environmental uncertainties and eliminate the need for labeled training data. Our extensive experimental analysis, conducted in a Private Payload Delivery Network (PPDN) —a common application in the autonomous multi-agent transportation domain—demonstrates that RAMPART successfully defends against a DP-exploited poisoning attack with a \(30\% \) attack ratio, achieving an F1 score of 0.852 and accuracy of \(96.3\% \) in heavy-traffic environments .

Individual-level activity-based modeling and indicators for assessing construction sites noise exposure in urban areas

Construction site noise significantly contributes to urban noise pollution. To accurately assess the noise exposure from urban construction sites, it is essential to consider the variability and dynamic nature of the urban context. This paper proposes an innovative approach using individual-level activity-based modeling to tackle the given challenge. The methodology introduced encompasses: the modeling of construction site noise emission sources, individual-level activity-based modeling of zone of influences, a modeling chain for assessing individual noise exposure levels, and a novel dedicated exposure indicator. To illustrate the relevance of our methodology, a case study is conducted in Champigny-sur-Marne, France; this study involves the generation of a synthetic population of 1,344,394 individuals distributed among 591,153 households, with 4,854,597 associated activities, using the Eqasim simulation framework. A traffic model is then established using the multi-agent transport simulation software MATSim. NoiseModelling software is employed to compute noise maps both with and without the construction site. Subsequently, the noise exposure for each individual is assessed throughout the day for both configurations. Thanks to this complete modeling chain, the proposed indicator quantifying the cumulative amount of time of exposure above a given threshold can be compute, with the case study result being estimated at more than 3,300 person.quarter-hours above 65 dB(A).

A Comparative Analysis of Time-Based and Hybrid Pricing Models for Electric Vehicle Charging

The accelerated adoption of Electric Vehicles (EVs) necessitates innovative and effective pricing strategies for charging infrastructure. This study leverages the MATSim (Multi-Agent Transport Simulation) framework to meticulously evaluate the performance of two distinctive EV charging pricing models: time-based charging and a hybrid model integrating both usage-based and time-based components. Driven by the pivotal question of how EV charging should be optimally priced—whether contingent on energy consumption or charging duration—the research endeavors to conduct two simulations. These simulations aim to provide a comprehensive comparative analysis, evaluating metrics such as total power derived from grids, utility revenue, charging station queues and served vehicles. In the time-based charging scenario, EVs incur charges based on their plugged-in duration, reflecting a pricing approach that correlates directly with the time a vehicle remains connected to the charging station. In contrast, in the hybrid model, EVs undergo initial billing based on usage until a predetermined battery charge point is reached, such as achieving a full battery. Subsequently, time-based pricing takes effect until the user disconnects the vehicle. The findings indicate that, in the combined approach, utility owners have the potential to generate more revenue. Conversely, the time-based approach demonstrates a capacity to serve a higher number of electric EVs, with comparable queue lengths observed in both approaches. Importantly, using the results of this paper, policymakers can suggest pricing schemes that maximize benefits for utility owners, reduce queues at charging stations, and ensure the security of power grids.

Multi-Agent Multimodal Transportation Simulation for Mega-cities: Application of Los Angeles

This paper presents a comprehensive modeling framework for large-scale agent-based multimodal transportation simulation tailored to the complex urban environments of megacities, with a specific focus on the Greater Los Angeles (LA) area. The framework leverages the Multi-Agent Transport Simulation (MATSim) toolkit, which addresses the evolving challenges posed by urbanization and technology-driven shifts in human mobility patterns. The framework enables the generation of explicit travel trajectories for a representative 10% sample population of LA County, comprising over one million individuals. To ensure the model’s validity, an efficient network calibration algorithm is proposed with rigorous validation using publicly available traffic count data. This paper underscores the development and configuration of the agent-based simulation model for LA, providing a robust foundation for large-scale urban mobility simulations and advancing our understanding of the intricacies of megacities.

Задача управления мультиагентной транспортной системой

A new class of control problems for a certain transport system is considered. The peculiarity of the system is the variable number of participants in the control process. The regrouping of participants, along with their unification and dispersal, is determined by the specified target points of visit, the initial point of departure and the final point of completion of the system’s movement, and the constructed trajectories. The structure of a multi-agent transport system implies the presence of primary and secondary carriers, with the former being able to transport the latter at some stages. It is at such moments that the number of participants in the control process changes. In the constructed mathematical model of the movements of the system under consideration and its control, a hybrid system of differential equations with switching is obtained, containing continuous and discrete components. The problem of finding optimal routes under certain constraints is solved.

Integration of Level 3 probabilistic risk assessment for nuclear power plants with transportation simulation considering earthquake hazards

This research improves the realism of Level 3 probabilistic risk assessment (PRA) for nuclear power plants (NPPs) to avoid subjective expert judgement by modeling the evacuation behavior of the residents and making two advancements. First, evacuation speed taken from a transportation simulation code was used as an input to the Level 3 PRA code, i.e., the MELCOR Accident Consequence Code System (MACCS). The transportation simulation code uses the Multiagent Transport Simulation (MATSim) evacuation extension software. MATSim is interfaced with MACCS, by generating input-output communication through evacuation routes and speed distributions. Integrating the agent-based transportation simulation with the Level 3 PRA code enables the explicit incorporation of spatiotemporal evacuation processes into NPP offsite risk assessment. Second, this research is the first to explicitly incorporate the probability of seismic damage to roadway bridges into Level 3 PRA of NPPs. The likelihood of seismic bridge damage, given an earthquake, is estimated using the HAZUS program developed by the Federal Emergency Management Agency (FEMA). The road closure due to seismic damage, estimated by HAZUS, is used as input to the agent-based transportation simulation, where evacuation routes and speed distributions are calculated considering the impact of seismic damage. These advancements can improve the realism of Level 3 PRA for NPPs by enhancing the evacuation analysis resolution and reducing the reliance on subjective judgment to characterize evacuation routes, speed, and seismic impact on Level 3 PRA. Integrating the transportation simulation and the natural hazard risk assessment with the Level 3 PRA code can also provide more in-depth causal insights for offsite risk management, such as recommending optimal evacuation routes in an emergency response plan. These contributions are demonstrated using the Sequoyah NPP case study adopted from the state-of-the-art reactor consequence analyses (SOARCA) conducted by the U.S. Nuclear Regulatory Commission (NRC).

The Multi-Agent Transportation Problem

We introduce the multi-agent transportation (MAT) problem, where agents have to transport containers from their starting positions to their designated goal positions. Movement takes place in a common environment where collisions between agents and between containers must be avoided. In contrast to other frameworks such as multi-agent pathfinding (MAPF) or multi-agent pickup and delivery (MAPD), the agents are allowed to separate from the containers at any time, which can reduce the makespan and also allows for plans in scenarios that are unsolvable otherwise. We present a complexity analysis establishing the problem's NP-completeness and show how the problem can be reduced to a sequence of SAT problems when optimizing for makespan. A MAT solver is empirically evaluated with regard to varying input characteristics and movement constraints and compared to a MAPD solver that utilizes conflict-based search (CBS).

Evaluation of Low-Traffic Neighborhoods and Scale Effects: The Paris Case Study

This paper investigates the socio-economic impacts of the driving restriction zone policy for the Paris region, using the multi-agent transport simulation MATSim. Based on a previous work which has calibrated a first instantiation of the mobility model to the Paris region, we extend it by modeling and simulating two forms of intermodality: Park-and-Walk and Park-and-Ride. After re-calibrating the model, we evaluate two policy scenarios featuring different scales of the restriction zone, using indicators related to mobility, traffic emissions, and social cost. The analysis puts special attention on how modal shift and emissions vary across space, that is, inside and outside the policy zone. We find that the small-scale scenario with a restriction zone limited to the Paris center leads to very limited modal shift, whereas the larger-scale policy scenario encompassing the whole of Paris shifts significantly more trips (about 1.4% of total trips) from cars to public transport and other modes. In both scenarios, traffic emissions decline in the restriction zone and its near surroundings. Far from the restriction zone, however, emissions increase in the small-scale scenario but decrease in the large-scale scenario. These differences are caused by the traffic detour, modal shift, and rebound effects, the magnitudes of which vary depending on the scale of the restriction zone. The driving restrictions also cause the social cost to increase by around 10% in both scenarios. This ultimately implies a trade-off between reducing emissions and increasing user costs, with a substantially greater efficiency (larger reduction for similar cost) in the large-scale policy scenario.

Operational Impacts of On-Demand Ride-Pooling Service Options in Birmingham, AL

Transportation Network Companies (TNCs) use online-enabled apps to provide on-demand transportation services. TNCs facilitate travelers to connect with drivers that can offer them rides for compensation using driver-owned vehicles. The ride requests can be for (a) individual or (b) shared rides. The latter, also known as ride-pooling services, accommodates requests of unrelated parties with origins and destinations along the same route who agree to share the same vehicle, usually at a discounted fare. Uber and Lyft offer ride-pooling services in select markets. Compared to individual ride requests, ride-pooling services hold better promise toward easing urban congestion by reducing the number of automobiles on the road. However, their impact on traffic operations is still not fully understood. Using Birmingham, AL as a case study, this research evaluated the impact that ride-pooling services have on traffic operations using a Multi-Agent Transport Simulation (MATSim) model of the Birmingham metro area. Scenarios were developed to simulate baseline conditions (no TNC service) and ride-pooling availability with two types of ride-pooling services, namely door-to-door (d2d) and stop-based (sB) service and three fleet sizes (200, 400, and 800 vehicles). The results indicate that when TNC vehicles are added to the network, the Vehicle Kilometers Traveled (VKT) decrease by up to 5.78% for the door-to-door (d2d) service, and up to 2.71% for stop-based (sB) services, as compared to the baseline scenario (no TNC service). The findings also suggest that an increase in the size of the ride-pooling fleet results in a rise in total ride-pooling service VKT, network-wide total VKT, and detour distance. However, increasing the size of the ride-pooling fleet also results in a decrease in the ride request rejection rates, thus benefiting the customers and decreasing the vehicle empty ratio which, in turn, benefits the TNC drivers. The results further suggest that a fleet of 200 ride-pooling vehicles can meet the current demand for service in the Birmingham region at all times, thus it is the optimal ride-pooling TNC fleet size for a medium-sized city such as Birmingham.

Analysis of cost-efficient urban air mobility systems: Optimization of operational and configurational fleet decisions

With the introduction of low-noise and low-emission electric vertical take-off and landing vehicles, passenger air transportation in urban areas is becoming increasingly important. Previous studies have designed vehicle concepts based on reference mission profiles, however, without considering strategic decisions about fleet operations, such as charging infrastructure, range and battery capacity. Comprehensive cost analyses for urban air mobility fleets have been largely neglected. In this paper, we develop an optimization model to analyze the cost-efficient operation of urban air mobility systems. Strategic decisions on vehicle concepts, battery capacity, and charging infrastructure are incorporated and evaluated using a total cost of ownership approach. We consider state-of-the-art modeling approaches, including vehicle-specific parameters, for accurate calculation of energy consumption. The optimization model is applied to the multi-agent transport simulation MATSim in a case study of the Ruhr (Germany) scenario to evaluate key parameter variations.The study results indicate the feasibility of an urban air mobility system in the study region with trip costs ranging from 27 US-$ to 46 US-$ per requested trip, depending on the scenario settings. Based on parameter variations, we find that a high cruising speed has a detrimental effect on the total cost. A medium charging power level is sufficient and energy costs account for only a moderate share. The latter is in contrast to previous research results, but can be attributed to the more detailed modeling approach of vehicle-specific parameters. We propose a cost incentive system as the ground handling time of the vehicles outweighs the recharging and flight time. The overall results provide a promising basis for model extensions.

Prioritization of Exigency Services in Multi-Agent Transportation Systems

Prioritization of Exigency Services in Multi-Agent Transportation Systems

The impact of dispatching logic on the efficiency of Urban Air Mobility operations

The application of electrified propulsion systems is expected to facilitate the future operation of small aircraft in urban and suburban areas. However, in addition to the safe design of vehicles, efficient operation must also be achieved by matching passenger requests to unoccupied and suitable vehicles. In the agent-based transportation simulation MATSim (Multi-Agent Transport Simulation), a rule-based approach has been followed by assigning the nearest available vehicle to each passenger request. Sophisticated considerations such as pooling multiple passengers with identical trip requests in terms of time and location or the urgency of a request were only partially considered or not considered at all. To decrease operating costs, especially by reducing the number of passenger flights and avoiding relocation flights, appropriate dispatching is essential for efficient assignment of available vehicles to trip requests. For this purpose, the suitability of the Bipartite Matching Dispatcher (BMD), already proven for ground-based on-demand mobility concepts, was investigated for application to UAM systems by demonstrating its performance on an existing data set using a mock-up simulation and comparing it to the initial dispatcher in MATSim. Studies have not only shown the similarity between UAM and ground-based on-demand operations, but also the compatibility of BMD for high load traffic scenarios with undersized fleets. Insights on the impact of available information, for example, the prioritization of early-submitted requests over urgent requests, on the operator’s point of view can be concluded. For an efficient vehicle assignment, the consideration of passengers’ attitude towards possible comfort restrictions due to longer waiting times and incentives through price adjustments is of crucial importance.

Tenerife’s Infrastructure Plan for Electromobility: A MATSim Evaluation

According to the Canarian government’s plans, a complete decarbonization of the Canary Islands economy is foreseen from 2040 onwards, which includes the electrification of land transport in the archipelago. However, due to the current low penetration rate of electric vehicles (EVs) on the islands, the number of EVs in circulation is expected to grow significantly in the coming years. Despite this, the network of charging points in Tenerife is currently totally insufficient, which is why it is essential to carry out a study to design the network of charging points in such a way that it can absorb the entire fleet of EVs that is expected to be in place by 2040. To this end, there are studies on the capacity, in terms of parking space, available for the installation of these charging points, but to date there are no studies on this subject supported by mobility data. For this reason, a simulation of traffic in Tenerife in 2040 has been carried out using MATSim (Multi-Agent Transport Simulation) to determine the ideal places to install these charging points and to find the number of charging points needed for the network.

Analysis of Charging Infrastructure for Private, Battery Electric Passenger Cars: Optimizing Spatial Distribution Using a Genetic Algorithm

To enable the deployment of battery electric vehicles (BEVs) as passenger cars in the private transport sector, suitable charging infrastructure is crucial. In this paper, a methodology for the efficient spatial distribution of charging infrastructure is evaluated by investigating a scenario with a 100% market penetration of BEVs of (around 1.3 million vehicles) in Berlin, Germany. The goal of the evaluated methodology is the development of various charging infrastructure scenarios—including public and private charging—which are suitable to cover the entire charging demand. Therefore, these scenarios are investigated in detail with a focus on the number of public charging points, their spatial distributions, the available charging power, and the necessary capital costs. For the creation of these charging infrastructure scenarios, a placement model is developed. As input, it uses the data of a multi-agent transport simulation (MATSim) scenario of the metropolitan area of Berlin to evaluate and optimize different distributions of charging infrastructure. The model uses a genetic algorithm and the principle of multi-objective optimization. The capital costs of the charging points and the mean detour car drivers must undertake are used as the optimization criteria. Using these criteria, we expect to generate cost-efficient infrastructure solutions that provide high usability at the same time. The main advantage of the method selected is that multiple optimal solutions with different characteristics can be found, and suitable solutions can be selected by subsequently using other criteria. Besides the generated charging scenarios for Berlin, the main goal of this paper is to provide a valid methodology, which is able to use the output data of an agent-based, microscopic transport simulation of an arbitrary city or area (or even real driving data) and calculate different suitable charging infrastructure scenarios regarding the different optimization criteria. This paper shows a possible application of this method and provides suggestions to improve the significance of the results in future works. The optimized charging infrastructure solutions for the Berlin scenario show capital costs of between EUR 624 and 2950 million. Users must cover an additional mean detour of 254 m to 590 m per charging process to reach an available charging point. According to the results, a suitable ratio between the charging points and vehicles is between 11:1 and 5:1. A share of fast charging infrastructure (>50 kW) of less than ten percent seems to be sufficient if it is situated at the main traffic routes and highly frequented places.

FABILUT: The Flexible Agent-Based Integrated Land Use/Transport Model

Integrated land-use transport models are often accused of being too complex, too coarse or too slow. We tightly couple the microscopic land use model SILO (Simple Integrated Land Use Orchestrator) with the agent-based transport simulation model MATSim (Multi-Agent Transport Simulation). The integration of the two models is person-centric. It means, firstly, that travel demand is generated microscopically. Secondly, SILO agents can query individualized travel information to search for housing or jobs (and to choose among available modes). Consequently, travel time matrices (skim matrices) are not needed anymore. Travel time queries can be done for any time of the day (instead of for one or few time periods), any x/y coordinate (instead of a limited number of zones) and take into account properties of the individual. This way, we avoid aggregation issues (e.g., large zones that disguise local differences) and we can account for individual constraints (e.g., nighttime workers who cannot commute by public transport for lack of service). Therefore, the behavior of agents is represented realistically, which allows us to simulate their reaction to novel policies (e.g., emission-class-based vehicle restrictions) and to extract system-wide effects. The model is applied in two study areas: a toy scenario and the metropolitan region of Munich. We simulate various transport and land use policies to test the model capabilities, including public transport extensions, zones restricted for private cars and land use development regulations. The results demonstrate that the increase of the model resolution and model expressiveness facilitates the simulation of such policies and the interpretation of the results.

A strategic scheme for partnership supply networks focusing on green multi-agent transportations: a game theory approach.

Competition to gain more market share leads to the development of a dynamic environment and the fulfillment of an improvement cycle in supply networks. In this paper, as the first attempt, two strategic periods based on the introduction and growth phases of supply networks are considered. Each of these periods has its own characteristics and develops different competitive structures. The allocation of market share between competing supply networks is carried out on the extent of their relationships with green orientations in the network. The ownership of supply sources in these networks is replaced by that of relationships (partnership supply) developed by some parent brands. The efforts of each parent brand towards establishing green supply relationships lead to a greater market share for it. Further to competition between the supply networks, co-petition within each network also results in the emergence of different virtual alliance structures. The performance history of each brand and its subsidiary partners in the first period is included in the second period's assessments and leads to the development or decline of the network. Another dimension of the paper lies in the development of multi-agent transportation arising from the green requirements. Two gaming-based heuristic algorithms are proposed for handling the decentralized decision-making structures in the multi-period multi-stage platform under discussion. To evaluate the taken approach, a real-world inspired case is taken into account. The numerical results prove the better performance of the proposed scheme towards green coverage of the network and, thereupon, increased market share.

Do road users play Nash Equilibrium? A comparison between Nash and Logistic stochastic Equilibriums for multiagent modeling of road user interactions in shared spaces

Analyzing and modeling multiagent transportation systems such as cyclist-pedestrian interactions and their evasive-action mechanisms in shared spaces are important for operation and safety evaluations. However, there are limited studies that (1) modeled the multiagent nature of road user interactions and their concurrent sequential decision processes, and (2) investigated the ability of different equilibrium behavioral theories in predicting road user operational-level decisions and evasive-action mechanisms. This study proposes two novel multiagent approaches based on different equilibrium theories for modeling road user interactions: (1) the Multiagent Generative Adversarial Imitation Learning (MAGAIL), which utilizes Nash-Equilibrium (NE) theory that assumes fully rational and optimal road user behavior, and (2) the Multiagent Adversarial Inverse Reinforcement Learning (MAAIRL), which utilizes Logistic-Stochastic-Best-Response-Equilibrium (LSBRE) theory that handles bounded rationality (sub-optimal) behavior. Unlike many of the traditional game-theoretic modeling approaches, which consider single time-step payoff modeling, the proposed approaches depend on Markov-Game that accounts for the stochastic nature of road user interactions and their sequential decision processes. The models recover road user multiagent reward functions and estimate their strategies using Multiagent Deep Reinforcement Learning. Using trajectories from three shared spaces in the USA and Canada, the study compared the proposed approaches’ results and determined a behavior-based consistent paradigm to model equilibrium in multiagent transportation systems. The results show that LSBRE-based model predicted road user trajectories and their evasive action mechanisms with higher accuracy compared to the NE-based model.

Agent-Based Simulation Approach to Determine Safety Impacts of Demand-Responsive Transport (DRT) in Wayne County, Michigan

Road safety is one of the major concerns in transportation system management. Safety performance analyses usually assess crash frequencies and the impacts of countermeasures on the number of crashes. However, the advent of new mobility solutions makes safety evaluation more challenging; data tend to be sparse, and the impacts of such services on demand and the performance of the broader network is not completely understood. This paper attempts to fill this gap by creating a novel method to estimate the safety impacts of Demand-Responsive Transport (DRT) services. Using an agent-based mesoscopic model in the Multi-Agent Transport Simulation (MATSim) Toolkit, we simulate DRT as a new mobility solution in Wayne County, Michigan, and obtain spatial and temporal distributions of traffic volume and operating speed to predict the frequency and severity of crashes. We tested for different DRT service designs, such as fleet size, detour tolerance, and initial placement of the fleet. Our findings indicate that introducing a DRT service on its own increases vehicle kilometers traveled (VKT) by 22% and consequently the number of crashes by 17%. However, this impact could be ameliorated by increasing the detour tolerance, which can significantly increase the number of shared rides and decreases the crash frequency (10.8%) and crash severity as a result of a lower VKT (15%). The output of the proposed framework serves as a starting point of a safety performance evaluation using an agent-based simulation approach to achieve a safe and successful implementation of shared mobility services in large-scale urbanized areas.