Multi-agent Transport Simulation Research Articles

Increased Convergence Rates in Multiagent Transport Simulations with Pseudosimulation

A multimodeling approach to large-scale, activity-based, multiagent simulation of travel demand is introduced. MATSIM is a full activity-based transport simulation. Its greatest current performance limitation is the network loading simulation, currently a queue simulation (QSim). QSim is iteratively executed for the entire agent population for evaluating the effects of random mutations on the activity plans of a fraction of the population. After each QSim, poorly performing plans are discarded, good plans are kept, and the agents slowly learn what works best for their individual activity needs. In the application presented, the system periodically replaces QSim for a number of iterations with a simplified pseudosimulation that runs approximately two orders of magnitude faster. The pseudosimulation uses travel time information from the preceding QSim iteration to estimate how well an agent day plan might perform. Repeated iterations of the pseudosimulation produce better-performing plans in a short time. These plans are passed to the QSim for updating of network travel time information, and the process repeats. The technique is tested in a scenario for Zurich, Switzerland, and incorporates mode choice, road pricing, secondary activity location choice, activity timing adjustment, and dynamic routing. The technique dramatically improves convergence rates for such complex, large-scale simulations and fully exploits modern multicore computer architectures.

Extensible Software Design of a Multi-Agent Transport Simulation

This paper explains the chosen methodology for software design of the Multi-Agent Transport Simulation, MATSim. The design focusses on standard architectures and design patterns to ease usability and improve extensibility of the software. Potential for extension is discussed using an example implementation of a traffic signal control module. Both, MATSim and the extension are using the same concepts for software architecture.

Analyzing Commercial Through-Traffic

The impact that commercial vehicles, especially freight, has on traffic is disproportionately large compared to the number of vehicles they represent. Instead of treating commercial vehicles and commercial through-traffic as back- ground noise in our transport planning models, this paper aims to shed light on the characteristics of commercial through-traffic. We argue that if we better understand the activities and activity chains, we would be in a position to build better and more realistic models that, in turn, will assist in better decision-making. In this paper we analyzed the activity chains of through-traffic, drawing from a pool of more than 30000 commercial vehicles that were tracked for six months. The results of the analysis show on an hour-by-hour basis where vehicles enter the study area, the number of activities conducted within the area, and the point of departure. In the porously bounded economic hub of South Africa, the province of Gauteng, we show that the majority of vehicles come into and leave the province through the same arterial routes. We also find distinguishing characteristics between through-traffic originating from within, versus those that originate from outside the province. It is a novel contribution that investigates the activity chain characteristics in a disaggregated manner, and lays a new foundation to build better transportation models in which freight traffic is reflected more accurately in an urban environment. Understanding the disaggregate activity chain structures allows us to generate synthetic populations of commercial vehicles and model them in an agent-based setting, such as the Multi-Agent Transport Simulation (MATSim) toolkit.

Within-Day Replanning of Exceptional Events

Typical software used in the simulation of traffic behavior focuses on scenarios describing common situations, such as an ordinary working day without remarkable incidents. To simulate such scenarios, iterative approaches are used. They assume that the people simulated adapt to previous iterations’ results. Such iterative approaches produce meaningful results for various scenarios only when typical, repetitive situations are modeled. However, a scenario may also contain incidents that occur randomly, and thus substantially increase a model's complexity. In such scenarios, an iterative approach would produce illogical and even erroneous results. Within-day replanning is an attempt to handle such scenarios. This paper describes problems that arise when an iterative simulation approach is applied to a scenario with exceptional events. The within-day replanning technique is introduced and implemented in the multiagent transport simulation framework, allowing simulated agents to replan the routes between their activities while they are traveling. By doing this, agents can take current traffic conditions into account, an important requirement for scenarios containing unpredictable incidents such as road accidents. The implementation capability is demonstrated by conducting experiments in which capacities of several arterial roads in the city center of Zurich, Switzerland, are reduced as the result of an exceptional event. It is demonstrated that agents affected by those events are able to reduce their travel times if they replan their routes by using within-day replanning.

Multi-Agent Transport Simulation Model for Eco-Commuting Promotion Planning

The global warming problem due to greenhouse gas emission has become very serious. Moreover, in the transportation section, the construction of a sustainable transportation system in the environment has been an important subject. Not only improvement of service level of modes, except those of private vehicles, but also consciousness for the environmental problem of an individual trip maker is important for eco-commuting promotion. Eco-consciousness can be changed under the influence of other people. The multi-agent simulation system for planning of ecocommuting promotion is developed using a social interaction model for eco-consciousness in this study. Modal shift is modeled by considering the result of a transport social experiment for eco-commuting promotion as reference. Influence of the promotion plan such as improvement of service level of public transport, promotion of individual eco-consciousness, and strengthening the interaction of individuals can be analyzed using the proposed model. Finally, it is concluded that the proposed multi-agent simulation with social interaction for eco-consciousness is useful for the planning of eco-commuting promotion.

Income-contingent user preferences in policy evaluation: application and discussion based on multi-agent transport simulations

Standard economic policy evaluation allows the realization of projects if the aggregated economic benefit outweighs their costs. The use of one single aggregated welfare measure for evaluating and ranking projects has often been criticized for many reasons. A major issue is that differentiated effects on individuals or subgroups of the population are not taken into consideration. This leads to the need for transport planning tools that provide additional information for politicians and decision makers. The microscopic multi-agent simulation approach presented in this paper is capable of helping to design better solutions in such situations. In particular, it is shown that the inclusion of individual income in utility calculations allows a better understanding of problems linked to public acceptance. First, individual income-contingent utility functions are estimated based on survey data in order to describe human mobility behavior. Subsequently, using the MATSim framework, the implementation is tested in a test scenario. Furthermore, and going beyond Franklin (2006), it is shown that the approach works in a large-scale real world example. Based on a hypothetical speed increase of public transit, effects on the welfare distribution of the population are discussed. It is shown that the identification of winners and losers seems to be quite robust. However, results indicate that a conversion or aggregation of individual utility changes for welfare analysis is highly dependent on the functional form of the utility functions as well as on the choice of the aggregation procedure.

A coupled multi-agent microsimulation of social interactions and transportation behavior

Choice set formation, location and mode preferences, coordinated scheduling, alternative utility valuations, and shared mobility resources are among the many activity-travel issues hypothesized to be significantly influenced by traveler interdependencies. Empirical evidence lags theory, particularly about the geography of social networks. A simulation tool is presented to let the experimenter construct and test hypothetical interdependencies between geography, socially-linked travelers, and activity-travel choices. The exploratory tool is integrated in the Multi-Agent Transportation Simulation Toolbox (MatSim-T). Initially, any social network can be constructed and embedded in geography. It can remain static, or be adapted to the travel patterns of the agents. The interactions and exchanges between agents influencing socializing and/or travel behavior can be defined in substance and in time/space. The reward for socializing or being socially linked can be varied. Finally, the co-dependence of social factors and travel behavior can be studied. This paper introduces the model and presents verification results which illustrate the coupling of extremely simplified socializing assumptions and travel behavior.

The representation and implementation of time-dependent inundation in large-scale microscopic evacuation simulations

Multi-agent simulation has increasingly been used for transportation simulation in recent years. With current techniques, it is possible to simulate systems consisting of several million agents. Such multi-agent simulations have been applied to whole cities and even large regions. In this paper it is demonstrated how to adapt an existing multi-agent transportation simulation framework to large-scale pedestrian evacuation simulation. The underlying flow model simulates the traffic-based on a simple queue model where only free speed, bottleneck capacities, and space constraints are taken into account. The queue simulation, albeit simple, captures the most important aspects of evacuations such as the congestion effects of bottlenecks and the time needed to evacuate the endangered area. In the case of an evacuation simulation the network has time-dependent attributes. For instance, large-scale inundations or conflagrations do not cover all the endangered area at once. These time-dependent attributes are modeled as network change events. Network change events are modifying link parameters at predefined points in time. The simulation framework is demonstrated through a case study for the Indonesian city of Padang, which faces a high risk of being inundated by a tsunami.

Network breakdown “at the edge of chaos” in multi-agent traffic simulations

Traffic is highly influenced by network structure and human behaviour. Small changes in the human behaviour can lead to huge changes in the load of a traffic network. Current transportation models do not, and most of them cannot, research such random behaviour but always calculate a steady state. In our multi-agent transport simulation, we frequently observe seemingly random "network breakdowns", huge traffic jams that spread over a big part of the network, making a normal traffic flow impossible. This paper describes the investigations that were performed on the results of our large-scale multi-agent transport simulations in an attempt to contribute to the better understanding of the dynamic processes in such simulations and, hopefully, better understanding and modelling of the real-world.

Multi-Agent Transport Simulations and Economic Evaluation

Summary Tolls are frequently discussed policies to reduce traffic in cities. However, road pricing measures are seldom implemented due to high investments and unpopularity. Transportation planning tools can support planning authorities by solving those problems if they take into account the following aspects: – Demographic attributes like income and time constraints – Time reactions to the policy – Schedule changes of population’s individuals during the whole day Our approach uses multi-agent simulations to model and simulate full daily plans. Each of our agents has a utility function that appraises the performance of a typical, microscopically simulated day. The sum of all utility changes to a policy change can be interpreted as the change in the system’s welfare thus the economic evaluation of a measure straightforward. The approach is tested with travel behavior of the Zurich metropolitan region in Switzerland. Several tolling schemes are investigated. It is shown that the simulation can be used to model travelers’ reactions to time-dependent tolls in a way most existing transportation planning tools are not able to do. It is demonstrated that route adjustment only, as is done in many traditional transport planning packages, results in no economic gains from the tolls. As time-dependent tolls are a much-debated subject in transportation politics, the ability to fully model such tolls and the reactions of travelers may help to find better toll schemes. In a world where individuals have more and more freedom to schedule their daily plans, agent-based simulations offer an intuitive way to research complex topics with lots of interdependencies.

An agent-based microsimulation model of Swiss travel: First results

In a multi-agent transportation simulation, each traveler is represented individually. Such a simulation consists of at least the following modules: (i) Activity generation. (ii) Modal and route choice. (iii) The traffic simulation itself. (iv) Learning and feedback. In order to find solutions which are consistent between the modules, a relaxation technique is used. This technique has similarities to day-to-day human learning. Using advanced computational methods, in particular parallel computing, it is now possible to run such a system for large metropolitan areas with 10 million inhabitants or more. This paper reports on such a simulation system for all of Switzerland. Our focus is on a computationally efficient implementation of the agent-based representation, which means that in fact each agent is represented with an individual set of plans as explained above. A database is used to store the agents' strategies, which are loaded into the simulation modules as required; the modules then feed back individual performance measures into the database. This approach allows that additional modules can be coupled easily, and without degrading computational performance. The set-up was tested for Swiss morning peak traffic. Hourly demand matrices were taken from work with the VISUM assignment package and converted to our needs. Routes were assigned via feedback learning using the agent data base. In other words, the current implementation uses a car-only versions of the modules (ii), (iii), and (iv). Resulting flow volumes are compared to the VISUM assignment results, and to field data.