100,000 robo-taxis driving in Berlin (10% sample)

MATSim – Mobility as a Service

Build Status

This project contains a collection of examples to simulate Mobility as a Service (MaaS) and Mobility on Demand (MoD) in MATSim. All services may be simulated with a driver or using Autonomous Vehicles (AVs). The basic framework for these services are the Dynamic Vehicle Routing Problem (DVRP), Autonomous Vehicles, Taxi and Demand Responsive Transport (DRT) extensions. This means, vehicles will be dispatched on-line while the MATSim Mobility Simulation is running.

Integration of DVRP into MATSim

The main goal of the code in this repository is to provide examples of different usage scenarios for MaaS / AV services and make them easy to access in one single place, while the actual optimizer code remains in the MATSim contributions. All the examples run on the current MATSim Snapshot, as there are continuous improvement to the functionality.


The following extensions might be of particular interest:


The centralized dispatched of Taxi services may be simulated using different algorithms, which may be set in the Config file. An overview of different Taxi dispatch strategies and their performance is provided in:

M. Maciejewski; J. Bischoff, & K. Nagel An Assignment-Based Approach to Efficient Real-Time City-Scale Taxi Dispatching, IEEE Intelligent Systems, 2016, 31, 68-77 Available here

Robotaxi / Shared Autonomous Vehicles

Shared Autonomous Vehicles provide the capability to simulate large fleets of automated taxi vehicles. This is done by combining a fast, rule-based dispatch algorithm with the possibility to adjust the consumed road capacity of Autonomous Vehicles.

The algorithm and results are presented in J. Bischoff, M. Maciejewski Simulation of city-wide replacement of private cars with autonomous taxis in Berlin, Procedia Computer Science, 2016, 83, https://doi.org/10.1016/j.procs.2016.04.121

The effects of different road capacity use are described in M. Maciejewski, J. Bischoff Congestion Effects of Autonomous Taxi Fleets; Transport, 2017, 0, 1-10, Full text available here

Demand Responsive Transport (DRT)

Demand Responsive Transport allows the pooling of several passengers into a single vehicle. Several constraints, such as maximum travel times and maximum waiting times, can be taken into account when organizing the vehicle dispatch. Please find a full documentation here.

Autonomous Vehicles

All MaaS extensions may be simulated with and without drivers. Arguably, the biggest influence of AV operations are road capacity and pricing.

Common infrastructure

With DVRP being the common base to all the modules described here, there is some common infrastructure all of the MaaS modules share:

Test scenarios

The scenarios folder contains several test scenarios. These are roughly derived from existing MATSim scenarios, but often depict only the excerpt with relevance to MaaS of the scenario.

How to use

1) Check out this project using your favorite git client or just download as a zip

2) Import the folder as a new Maven project to Eclipse (Import --> Maven --> Existing project) or intelliJ (New --> Module from existing sources --> Select the folder --> Maven) 3) Run the example classes and start editing them according to your taste.

Using the Maven approach is strongly recommended.