MengeROS

MengeROS is a novel open-source crowd simulation tool for robot navigation that integrates Menge with ROS. It extends Menge to introduce one or more robot agents into a crowd of pedestrians. Each robot agent is controlled by external ROS-compatible controllers. MengeROS has been used to simulate crowds with up to 1000 pedestrians and 20 robots.

Installation
Robot Configuration
Crowd Configuration
Examples
Reference Paper
Contact
Acknowledgments

Installation

(Tested For Ubuntu 14.04 with ROS Indigo)

Download the source code into the catkin workspace of ROS.
Use catkin_make to recompile the ROS packages and then start roscore.
Run the simulator using the command:

rosrun menge_sim menge_sim -p examples\core\tradeshow.xml

The previous command should start the simulator with the tradeshow world and as many robots as specified in the examples\core\tradeshow\tradeshowS.xml. The various sample environments files included in MengeROS can be found in the examples folder in this repository.
Press space to start the simulation; pressing space again will pause the simulation.
The robot in the simulator can be controlled externally using any controller than sends Twist messages.

Robot Configuration:

To introduce a new robot agent into MengeROS, update the specification file (tradeshowS.xml). Create a new AgentProfile and replace

\<Common class="2" r="0.26" external="0"/>

with

\<Common class="2" r="0.26" external="1" start_angle="-1.96" end_angle="1.918" increment="0.0005817" range_max="25"/>.

The "external" variables ensure that the new agent is controlled externally. For more information on how to create an AgentProfile refer to the Menge documentation.

The radius of the robot can be configured by varying r=0.26.
The laser range, the field of view and the number of ray traces can be set by changing start_angle, end_angle, increment and range_max (See ROS documentation for more details). The configuration currently defaults to Fetch robot specifications.
All distances are in meters and angles are in radians.

Crowd Configuration:

The integration allows all Menge crowd configurations to be generated in MengeROS with no additional changes. The documentation on how to generate various crowd scenarios is given on the Menge website.

Examples:

Here are several examples of MengeROS simulations.

This video shows the tradeshow world in action with a single robot that is controlled by an external controller.
This video compares ORCA and the original social forces model (Helbing, Dirk, and Peter Molnar. "Social Force Model for Pedestrian Dynamics." Physical Review E 51.5 (1995): 4282.). Other examples of different collision avoidance models for pedestrians can be found here.
This video shows an application of MengeROS for crowd-sensitive path planning. In the video, a robot (blue) navigates through a simple office-like environment around 90 pedestrians. The robot learns a distribution of the crowd using only local sensor observations. The left side shows the simulator; the right side shows the rviz visualization where the laser endpoints are in red. The dark regions of the grid indicate the likelihood of dense crowds.

Reference paper

A paper describing MengeROS can be downloaded here. If you are going to use this library for your work, please cite it within any resulting publication:

A. Aroor, S.L. Epstein, R. Korpan "MengeROS: a Crowd Simulation Tool for Autonomous Robot Navigation", AAAI 2017 Fall Symposium on Artificial Intelligence for Human-Robot Interaction, 2017.

The bibtex code for including this citation is provided:

@INPROCEEDINGS{aroor2017,
  AUTHOR={Aroor, Anoop  and  Epstein, Susan L  and  Korpan, Raj},
  TITLE={MengeROS: A Crowd Simulation Tool for Autonomous Robot Navigation},
  BOOKTITLE={AAAI 2017 Fall Symposium on Artificial Intelligence for Human-Robot Interaction},
  YEAR={2017}}

Contact

Anoop Aroor The Graduate Center of The City University of New York aaroor@gradcenter.cuny.edu

Acknowledgments

The development of MengeROS was supported in part by NSF Grant #1625843 and by the Machine Learning and Problem Solving Lab at Hunter College, CUNY.

flztiii / menge_ros