Michal Maciejewski

Towards multi-agent simulation of the dynamic vehicle routing problem in MATSim

The paper presents the idea and the initial outcomes of integrating MATSim, a multi-agent transport simulation system, with the DVRP Optimizer, an application for solving the Dynamic Vehicle Routing Problem. First, the justification for the research is given and the state of the art is outlined. Then, MATSim is presented with a short description of the recent results in areas related to the one proposed in the paper, followed up by the discussion on the DVRP Optimizer functionality, architecture and implemented algorithms. Next, the process of integrating MATSim and the DVRP Optimizer is presented, with the distinction of two phases, the off-line and on-line optimization. Then, a description of the off-line optimization is given along with the results obtained for courier and taxi services in urban areas. The paper ends with conclusions and future plans.

Simulation of City-wide Replacement of Private Cars with Autonomous Taxis in Berlin

Abstract Autonomous taxi (AT) fleets have the potential to take over a significant amount of traffic handled nowadays by conventionally driven vehicles (CDV). In this paper, we simulate a city-wide replacement of private cars with AT fleets of various sizes. The simulation model comprises microscopic demand for all private car trips in Berlin (including incoming and outgoing traffic), out of which the internal ones are exclusively served by ATs. The proposed real-time AT dispatching algorithm was optimized to handle hundreds of thousands of vehicles and millions of requests at low computing times. Simulation results suggest, that a fleet of 100 000 vehicles will be enough to replace the car fleet in Berlin at a high service quality for customers. Based on this, one AT could replace the demand served by ten CDVs in Berlin.

An Assignment-Based Approach to Efficient Real-Time City-Scale Taxi Dispatching

This study proposes and evaluates an efficient real-time taxi dispatching strategy that solves the linear assignment problem to find a globally optimal taxi-to-request assignment at each decision epoch. The authors compare the assignment-based strategy with two popular rule-based strategies. They evaluate dispatching strategies in detail in the city of Berlin and the neighboring region of Brandenburg using the microscopic large-scale MATSim simulator. The assignment-based strategy produced better results for both drivers (less idle driving) and passengers (less waiting). However, computing the assignments for thousands of taxis in a huge road network turned out to be computationally demanding. Certain adaptations pertaining to the cost matrix calculation were necessary to increase the computational efficiency and assure real-time responsiveness.

Large-scale Microscopic Simulation of Taxi Services

The paper presents research on large-scale microscopic simulation of taxi services in Berlin based on floating car data collected by the Taxi Berlin fleet, the largest taxi association in Germanys capital. Firstly, Berlins taxi market is shortly described and the demand and supply data obtained from FCD analysed. Secondly, the online taxi dispatching problem formulation for this specific case is given, followed by the definition of two real-time rule-based heuristics used to dispatch taxis dynamically within the simulation. Finally, the simulation setup in MATSim is described, and the results obtained with both heuristics are analysed and compared in terms of dispatching performance, proving the effectiveness of the second strategy at different demand scales.

Analysis of Berlin's taxi services by exploring GPS traces

With current on-board GPS devices a lot of data is being collected while operating taxis. This paper focuses on analysing travel behaviour and vehicle supply of the Berlin taxi market using floating car data (FCD) for one week each in 2013 and 2014. The data suggests that there is generally a demand peak on workday mornings and a second peak over a longer time in the afternoon. On weekends, the demand peaks shift towards the night. On the supply side, drivers seem to adapt to the demand peaks very efficiently, with fewer taxis being available at times of low demand, such as during midday. A spatial analysis shows that most taxi trips take place either within the city centre or from/to Tegel Airport, the citys largest single origin and destination. Drivers spend a large amount of their work time on waiting for customers and the taxi rank at Tegel Airport is the most popular one.

Large-scale microscopic simulation of taxi services. Berlin and Barcelona case studies

The paper presents research on large-scale microscopic simulation of taxi services in Berlin and Barcelona based on floating car data collected by local taxi fleets. Firstly, Berlin’s and Barcelona’s taxi markets are shortly described and the demand and supply data obtained from FCD analysed. Secondly, the online taxi dispatching problem formulation for this specific case is given, followed by the definition of two real-time rule-based heuristics used to dispatch taxis dynamically within the simulation. Finally, the simulation setup in MATSim is described, and the results obtained with both heuristics are analysed and compared in terms of dispatching performance, proving the effectiveness of the second strategy at different demand and supply scales. This paper is an extended version of Maciejewski and Bischoff 2015, where only the Berlin case study was presented.

Benchmarking minimum passenger waiting time in online taxi dispatching with exact offline optimization methods

This paper analyses the use of exact offline optimization methods for benchmarking online taxi dispatching strategies where the objective is to minimize the total passenger waiting time. First, a general framework for simulating dynamic transport services in MATSim (Multi-Agent Transport Simulation) is described. Next, the model of online taxi dispatching is defined, followed by a formulation of the offline problem as a mixed integer programming problem. Three benchmarks based on the offline problem are presented and compared to two simple heuristic strategies and a hypothetical simulation with teleportation of idle taxis. The benchmarks are evaluated and compared using the simulation scenario of taxi services in the city of Mielec. The obtained (approximate) lower and upper bounds for the minimum total passenger waiting time indicate directions for further research.

The Influence of Multi-agent Cooperation on the Efficiency of Taxi Dispatching

The paper deals with the problem of the optimal collaboration scheme in taxi dispatching between customers, taxi drivers and the dispatcher. The authors propose three strategies that differ by the amount of information exchanged between agents and the intensity of cooperation between taxi drivers and the dispatcher. The strategies are evaluated by means of a microscopic multi-agent transport simulator (MATSim) coupled with a dynamic vehicle routing optimizer (DVRP Optimizer), which allows to realistically simulate dynamic taxi services as one of several different transport means, all embedded into a realistic environment. The evaluation is carried out on a scenario of the Polish city of Mielec. The results obtained prove that the cooperation between the dispatcher and taxi drivers is of the utmost importance, while the customer–dispatcher communication may be reduced to minimum and compensated by the use of more sophisticated dispatching strategies, thereby not affecting the quality of service.

Towards a Testbed for Dynamic Vehicle Routing Algorithms

Since modern transport services are becoming more flexible, demand-responsive, and energy/cost efficient, there is a growing demand for large-scale microscopic simulation platforms in order to test sophisticated routing algorithms. Such platforms have to simulate in detail, not only the dynamically changing demand and supply of the relevant service, but also traffic flow and other relevant transport services. This paper presents the DVRP extension to the open-source MATSim simulator. The extension is designed to be highly general and customizable to simulate a wide range of dynamic rich vehicle routing problems. The extension allows plugging in of various algorithms that are responsible for continuous re-optimisation of routes in response to changes in the system. The DVRP extension has been used in many research and commercial projects dealing with simulation of electric and autonomous taxis, demand-responsive transport, personal rapid transport, free-floating car sharing and parking search.

The Implementation of the Electric Taxi Fleet in the City of Poznan, Poland

Transport is an essential element of socio-economic growth, however, its development, besides the obvious benefits, generates a number of external costs such as congestion, noise, emission of harmful substances into the environment, depletion of non-renewable resources and energy consumption. One of the factors affecting the reduction of the environmental burden from road transport is the replacement of the conventionally-fueled vehicles by other types of powertrains, including the electric drives. The introduction of electric vehicles may become the key element in the sustainable transport strategy development in urban areas, but due to the fact that electric vehicles are still unpopular, the demand for these vehicles should be boosted by the vehicle fleet owners, especially by the companies providing transport services. The introduction of electric taxis requires the construction of an appropriate charging infrastructure, i.e. the selection of the charging station type and the determination of the number and locations of the stations. The paper outlines the progress in the introduction of electric taxi fleets in several cities. The main section relates to the possibility of the implementation of electric taxicabs in the city of Poznan, an agglomeration of 600 thousand inhabitants, located in the western part of Poland. The choice of vehicle and the calculation of the operating costs of a taxicab, as well as the calculation of the charging infrastructure costs have been also presented.