mod-abm-2.0

A modern agent-based modeling platform for mobility-on-demand simulations.

What's mod-abm-2.0?

The mod-abm-2.0 is the 2.0 version of our Agent-Based Modeling platform designed for the simulation of large-scale Mobility-on-Demand operations. The software largely extends and rewrites its original version (amod-abm), which was first developed in 2016 as part of an MIT research project. In this 2.0 version, we have introduced a lot of new designs and upgrades in pursuit of better scalability and extensibility. This includes:

• implementation in C++ with native OSRM instead of Python to allow for much optimized runtime performance;
• user-customizable simulations through hierarchical system configurations, decoupled routing/demand-generating modules and easily transferrable maps to support different use cases and research topics;
• thorough unit tests, conprehensive documentation and modern software development practices.

Almost effortlessly, mod-abm-2.0 can be set up to simulate city-level mobility-on-demand systems in any urban settings. It creates a free-floating MoD system with a fleet of vehicles and a central dispatcher, and models each of the agents (travelers, vehicles, couriers etc.) at the individual level. The simulation evaluates the system performance based on a set of user-defined indicators, such as wait time, travel time, service rate at the traveler side, and vehicle miles traveled, average load, vehicle idle time at the operator side. Additional Python tools are also provided to create animation from the simulated data for debugging/visualization purposes.

With all great features aforementioned, mod-abm-2.0 creates opportunities for researchers and planners to:

• support design and operational decisions (such as fleet size, pricing policies) for MoD systems, shared or non-shared, autonomous or non-autonomous;
• model traffic demand and understand travel behavior in interaction with MoD systems;
• evaluate smart dispatch algorithms w.r.t trip-vehicle assignment, fleet management, and rebalancing, including machine-learning models.

Thanks for contributing to mod-abm-2.0!

mod-abm-2.0 Explained

The simulation Platform in mod-abm-2.0 runs in conjunction with a Router and a DemandGenerator. In each cycle, Platform invokes DemandGenerator to generate trip requests based on the demand model. It then dispatches the trips to vehicles following the selected dispatching strategies. Router , which wraps around the OSRM backend, provides the best routes for trips and routing requests during dispatching. Once dispatched, each individual vehicle will navigate itself through the planned waypoints to serve the trips sequentially. Platform accumuates the states along the process, which then helps understand the simulated MoD system from perspectives like fleet usage, level of service etc.

The program outputs simulation results in two formats:

• a report as a quick summary of the simulation results (through terminal);
• a detailed datalog for debugging, visualization and in-depth investigation (in .yml format).

We also provide Python tools to parse the output datalog and create animation video in .mp4 format. Example of the rendered video can be found at demo.mp4.

As of today, mod-abm-2.0 has the following features developped or planned:

• Dispatching
  • Insertion Heuristics, which assigns trips to vehicles while minimizes the total travel time of all travelers;
  • [WIP, High Priority] interface with machine learning libraries for ML-based dispatching;
  • [TBD, Mid Priority] re-optimization through metaheuristics for better trip-vehicle matching;
  • [TBD, Low Priority] rebalancing of idle vehicles for better level of service;
• Router
  • OSRM routing engine on static local map data;
  • [TBD, Low Priority] routing with dynamic traffic data or historic traffic data;
• Demand Generator
  • time-invariant demand (that generates trips following Poisson process);
  • [TBD, High Priority] mode choice model that competes with transit and other modes of transportation;
• Others
  • [TBD, Mid Priority] simulation of transit systems for multi-modal interaction.

Full documentation of mod-abm-2.0 can be found below.

Ducumentation

• Quick Start to get started with mod-abm-2.0 on your machine.
• Runbook to understand how everything works under the hood and how to set up your own scenarios.
• FAQ for frequently asked questions.

Support

You can post bug reports and feature requests in Issues.

Relavant Papers

• Wen, J., Chen, Y.X., Nassir, N. and Zhao, J., 2018. Transit-oriented autonomous vehicle operation with integrated demand-supply interaction. Transportation Research Part C: Emerging Technologies, 97, pp.216-234. (pdf)

• Wen, J., Zhao, J. and Jaillet, P., 2017, October. Rebalancing shared mobility-on-demand systems: A reinforcement learning approach. 2017 IEEE 20th International Conference on Intelligent Transportation Systems (ITSC) (pp. 220-225). Ieee. (pdf)

• Wen, J., Nassir, N. and Zhao, J., 2019. Value of demand information in autonomous mobility-on-demand systems. Transportation Research Part A: Policy and Practice, 121, pp.346-359. (pdf)

Citing

For research work/publications that uses mod-abm-2.0, please cite as follows:

@misc{wen2021-mod-abm-2.0,
    author = {Wen, Jian},
    title = {mod-abm-2.0, an agent-based modeling platform for mobility-on-demand simulations},
    year = {2021},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\url{https://https://github.com/wenjian0202/mod-abm-2.0}},
}